Capillary Exchange
Every minute, roughly 20 liters of fluid filter out of your capillaries into surrounding tissue. Yet you do not swell up. Almost all of that fluid returns to the blood. It goes either directly across the capillary wall or via the lymphatic system. This balancing act — filtration versus reabsorption, drainage versus retention — is the subject of capillary exchange. Capillaries are the only blood vessels built for exchange rather than transport. Their walls are often a single squamous endothelial cell thick. They separate blood from interstitial fluid by less than a micrometer. Across that thin barrier, oxygen, carbon dioxide, glucose, hormones, and water move constantly between plasma and tissue. Understanding how this traffic is regulated explains several things. It explains why a burn victim goes into shock, and why liver disease causes ascites. It explains why standing still for hours makes ankles swell. It also explains why a low-protein diet produces edema even when blood volume is normal. The central conceptual tool is the Starling principle, formalized by Ernest Starling in 1896. Fluid movement across a capillary wall depends on the balance of hydrostatic and osmotic pressures on each side. This is not a vague balance of "pressures" in the abstract. It is a quantitative equation with four measurable terms. Pathology can be traced to a change in any one of them. Building on Hemodynamics & Blood Flow and Blood Pressure & Its Regulation, this article works through the mechanism. It covers the classic evidence and the clinical consequences of capillary fluid exchange. It then points toward Gas Exchange & Diffusion, where the same capillary walls handle a different kind of traffic. This article delivers several things. It covers the physical mechanism of diffusion and bulk flow across capillary walls. It explains the Starling equation term by term. It describes the lymphatic system's essential backup role. It presents the classic experiments that established the theory, along with worked numerical examples. Finally, it covers the major disease states that arise when the balance fails.
- Capillary walls exchange solutes by diffusion (fast, for small lipid-soluble and small polar molecules) and exchange fluid by bulk flow (filtration and reabsorption) through water-filled clefts and pores.
- Fluid movement is governed by the Starling equation: net filtration = capillary hydrostatic pressure + interstitial osmotic pressure − interstitial hydrostatic pressure − plasma osmotic (oncotic) pressure, scaled by filtration coefficient.
- Net filtration slightly exceeds net reabsorption across the whole systemic circulation, producing roughly 2–4 liters of excess interstitial fluid per day.
- The lymphatic system collects this excess fluid and protein and returns it to the venous circulation, preventing edema.
- Edema results from any imbalance: elevated capillary hydrostatic pressure, reduced plasma oncotic pressure, increased capillary permeability, or lymphatic blockage.
- Starling's own experiments and later microperfusion studies (Landis, Michel, Levick) confirmed and then refined the model, especially the discovery that reabsorption is far less complete than classically taught.
Fluid crosses capillary walls whenever the outward push of blood pressure and inward pull of interstitial protein exceed the inward pull of plasma protein and outward push of tissue pressure — and the lymphatics quietly return whatever slips out.
The core idea
A capillary is a tube of endothelial cells roughly 5–10 μm in diameter, thin enough that red blood cells must sometimes deform to pass through single file. Two fundamentally different transport processes occur across this wall. Diffusion moves individual solutes — oxygen, carbon dioxide, glucose, amino acids, ions — down their concentration gradients through the endothelial cells themselves (for lipid-soluble substances like O₂ and CO₂) or through water-filled intercellular clefts and fenestrations (for small polar solutes like glucose, ions, and water). Diffusion is extremely fast over the short distances involved and is the dominant mechanism for nutrient and gas exchange. Bulk flow, in contrast, moves whole volumes of protein-poor fluid across the wall through the same clefts, driven not by concentration gradients but by pressure differences — this is filtration and reabsorption, and it is the mechanism responsible for distributing extracellular fluid between the vascular and interstitial compartments. The pressures that drive bulk flow are collectively called Starling forces, after the English physiologist Ernest Starling. There are four:
| Force | Symbol | Direction | Typical value |
|---|---|---|---|
| Capillary hydrostatic pressure | P_c | pushes fluid out of capillary | ~25–35 mmHg (arteriolar end) to ~10–15 mmHg (venous end) |
| Interstitial hydrostatic pressure | P_i | pushes fluid into capillary (usually slightly negative) | ~ −3 to 0 mmHg |
| Plasma colloid osmotic (oncotic) pressure | π_p | pulls fluid into capillary | ~25–28 mmHg |
| Interstitial fluid colloid osmotic pressure | π_i | pulls fluid out of capillary | ~3–8 mmHg (tissue-dependent) |
Oncotic pressure is the osmotic pressure generated specifically by plasma proteins (chiefly albumin, which is small, abundant, and negatively charged) that cannot cross the capillary wall in significant amounts. Because these proteins are largely retained in plasma, they create an osmotic gradient that opposes filtration — this is distinct from total osmotic pressure, which is dominated by small ions and is nearly equal on both sides of the wall (and therefore does not drive net fluid movement).
Mechanism
Step 1 — Structural basis of exchange. Capillary walls come in three histological types that set the ceiling on what can cross. Continuous capillaries (muscle, skin, lung) have tight junctions with narrow intercellular clefts (~4 nm) — permeable to water, ions, and small solutes, but not proteins. Fenestrated capillaries (kidney glomeruli, intestinal villi, endocrine glands) have pores (~ 60–80 nm) spanned by a thin diaphragm, allowing much higher fluid and small-solute flux. Sinusoidal (discontinuous) capillaries (liver, spleen, bone marrow) have large gaps and incomplete basement membrane, permitting even plasma proteins and cells to pass. This is why liver sinusoids allow albumin to equilibrate almost freely between plasma and interstitium — a key reason liver disease causes edema partly through altered local Starling balance, not just through reduced albumin synthesis.
Step 2 — Diffusion of solutes. Lipid-soluble gases (O₂, CO₂) diffuse directly across the endothelial cell membranes, driven by partial pressure gradients, at rates so fast that equilibration occurs within a fraction of the capillary's length under resting conditions. Small polar solutes (glucose, amino acids, ions, water) diffuse through the intercellular clefts. Diffusion rate depends on the concentration gradient, the surface area available, the diffusion distance, and the permeability of the pathway (formalized in Gas Exchange & Diffusion via Fick's law). Diffusive flux across capillaries is enormous relative to bulk flow — for a substance like oxygen, total diffusional exchange across the systemic capillary bed each minute vastly exceeds the volume of fluid filtered, because diffusion does not require net fluid movement at all.
Step 3 — Bulk flow and the Starling equation. Net fluid movement (Jᵥ) across a capillary segment is described by:
Jᵥ = Kf × [(P_c − P_i) − σ(π_p − π_i)] where Kf is the filtration coefficient (a product of hydraulic conductivity and surface area — high in fenestrated/sinusoidal beds, low in continuous ones), and σ is the reflection coefficient (0 to 1), representing how effectively the wall excludes protein (σ ≈ 1 for a wall fully impermeable to protein; σ ≈ 0 for a wall that lets protein pass freely, eliminating its osmotic effect). A positive Jᵥ means net filtration (fluid leaving the capillary); negative means net reabsorption (fluid entering). At the arteriolar end of a typical systemic capillary, P_c (~30–35 mmHg) is high because it is closer to arterial pressure, so net outward force dominates and filtration occurs. At the venous end, P_c has fallen (~10–15 mmHg) due to resistance along the capillary, so the balance may shift toward net reabsorption, since π_p (~25 mmHg) now exceeds the reduced P_c. Classically this was taught as a clean symmetric picture: filtration at the arteriolar end, reabsorption at the venous end, net zero. Direct microperfusion measurements (see Evidence) have shown this simple picture is an oversimplification — in most continuous-capillary beds, filtration occurs along nearly the entire length, with net reabsorption being the exception rather than the rule, because the sub-glycocalyx protein concentration near the wall is lower than bulk interstitial values, keeping the effective π_i low and effective σ high everywhere.
Step 4 — The glycocalyx. A thin layer of membrane-bound proteoglycans and glycoproteins (the endothelial glycocalyx) coats the luminal surface of the endothelium and acts as the actual molecular sieve, with the interendothelial cleft junctions beneath it. Fluid that filters through the glycocalyx into the initial cleft carries very little protein, so the local oncotic gradient opposing filtration is set by the sub-glycocalyx protein concentration, not by bulk interstitial fluid — this "revised Starling principle" explains why sustained net reabsorption is rare under steady-state conditions, and why filtered fluid is returned mainly via the lymphatics rather than by venous-end reabsorption.
Step 5 — Lymphatic return. Across the whole body, net filtration modestly exceeds net reabsorption, producing an excess of roughly 2–4 liters of protein-containing fluid entering the interstitium per day beyond what capillaries reabsorb. This excess, along with any protein that leaked out, is picked up by blind-ended lymphatic capillaries, which have overlapping endothelial cells acting as one-way flap valves: interstitial pressure pushes fluid in, and back-flow is prevented by the flaps closing. Lymph is propelled by skeletal muscle contraction, smooth muscle contraction of larger lymphatic vessels, one-way valves, and respiratory pressure changes, eventually draining into the subclavian veins via the thoracic duct and right lymphatic duct. This closes the exchange loop: fluid and protein that leave the blood in capillaries are returned to the blood via an entirely separate vascular network.
The evidence
Ernest Starling's original experiments (1896). Working at University College London, Starling perfused isolated dog limbs and studied lymph formation while manipulating venous pressure and plasma protein concentration. He observed that raising venous (and hence capillary) pressure increased lymph flow from the tissue — direct evidence that hydrostatic pressure drives fluid out of vessels. Critically, he also showed that infusing protein-free saline into the circulation, which diluted plasma proteins, caused fluid to accumulate in tissues (edema) even without any change in hydrostatic pressure — demonstrating that plasma proteins exert an osmotic pull opposing filtration. Conversely, infusing concentrated protein solution drew fluid into the vasculature from the tissues. From these observations he proposed that capillary fluid exchange is governed by the balance of hydrostatic and "colloidal osmotic" pressures — the qualitative core of what is now called the Starling equation. Starling did not have the tools to measure interstitial hydrostatic or osmotic pressure precisely; those terms (P_i and π_i) were added and quantified only decades later.
Landis's direct microperfusion (1927). Eugene Landis, working with individual capillaries in frog mesentery under the microscope, cannulated single capillaries and measured filtration and reabsorption rates directly as a function of imposed hydrostatic pressure, confirming a near-linear relationship between capillary pressure and fluid movement, and providing the first direct experimental confirmation of Starling's model at the level of a single vessel — a landmark in quantitative microvascular physiology.
Pappenheimer and Soto-Rivera, and later refinements. In the mid-20th century, whole-organ perfusion studies (using isolated perfused limbs and measuring capillary filtration coefficients) extended Starling's framework to estimate Kf for different tissues, showing large variation consistent with the fenestrated/continuous/sinusoidal structural classification.
Michel and Levick's revision (1990s–2000s). Using more precise microperfusion techniques on single perfused capillaries (notably in frog and rat mesentery), Charles Michel and J. Rodney Levick found that sustained net reabsorption predicted by the classical symmetric Starling picture rarely occurred in steady state — instead, most capillary beds showed continuous filtration, with the small amount of "reabsorption" occurring only transiently, e.g. after a hemorrhage lowers capillary pressure abruptly. This led to the glycocalyx model (or "revised Starling principle"), which holds that the relevant π_i for filtration calculations is the protein concentration in the thin sub-glycocalyx fluid layer, not the bulk interstitial fluid, which better predicted observed filtration and clarified the near-exclusive role of the lymphatics in returning filtered fluid, rather than venous-end reabsorption alone. This refinement is now the physiological consensus, though the classical Starling equation remains an excellent first-approximation teaching model.
Worked examples
Example 1 — Calculating net filtration pressure and predicting edema risk. A patient with nephrotic syndrome has lost large amounts of albumin in urine. Measured values:
| Variable | Normal | Patient |
|---|---|---|
| P_c | 30 mmHg (arteriolar end) | 30 mmHg (unchanged) |
| P_i | −2 mmHg | −2 mmHg (unchanged) |
| π_p | 26 mmHg | 12 mmHg (hypoalbuminemia) |
| π_i | 5 mmHg | 5 mmHg (unchanged) |
Net filtration pressure (NFP) = (P_c − P_i) − (π_p − π_i) Normal: NFP = (30 − (−2)) − (26 − 5) = 32 − 21 = +11 mmHg (net filtration, physiological — matched by lymphatic drainage) Patient: NFP = (30 − (−2)) − (12 − 5) = 32 − 7 = +25 mmHg The patient's net outward filtration pressure has more than doubled. Even with fully functional lymphatics, the lymphatic system's maximum drainage capacity (roughly 10× resting lymph flow) is eventually overwhelmed, and fluid accumulates in the interstitium as edema. This is why hypoalbuminemia (from nephrotic syndrome, liver failure, or malnutrition/kwashiorkor) produces generalized edema even when capillary hydrostatic pressure and cardiac function are entirely normal — the defect is purely in the oncotic term of the Starling equation.
Example 2 — Right heart failure and dependent edema. A patient with right-sided heart failure has elevated central venous pressure, which backs up into systemic veins and raises venous pressure throughout the body, especially in gravity-dependent regions (ankles, feet). Suppose in the lower limb capillaries:
- Normal P_c (venous end) = 15 mmHg → after right heart failure, P_c (venous end) rises to 28 mmHg (elevated venous back-pressure transmitted through capillaries).
- π_p unchanged at 26 mmHg, π_i unchanged at 5 mmHg, P_i unchanged at −2 mmHg.
Normal venous-end NFP = (15 − (−2)) − (26 − 5) = 17 − 21 = −4 mmHg (net reabsorption favored, per classical model) Heart failure venous-end NFP = (28 − (−2)) − (26 − 5) = 30 − 21 = +9 mmHg (filtration now favored even at the venous end) Because the venous end no longer reabsorbs fluid — and may even filter — total capillary filtration across the whole vessel rises sharply, lymphatic capacity is exceeded, and dependent (gravity-driven) pitting edema appears in the ankles, worsening by the end of the day as venous pressure in the legs is further elevated by prolonged standing. This illustrates why diuretics (reducing blood/venous volume, lowering P_c) and leg elevation (reducing gravitational venous pressure) are first-line treatments, while albumin infusion is comparatively less useful here, since the primary defect is hydrostatic, not oncotic.
Common misconceptions & traps
| Misconception | Correction |
|---|---|
| "Reabsorption exactly balances filtration at the venous end, so no net fluid is left over." | Filtration modestly exceeds reabsorption everywhere; the surplus (2–4 L/day) is normally removed by the lymphatics, not by perfect Starling balance. |
| "Oncotic pressure and total osmotic pressure are the same thing." | Total osmotic pressure (mostly from ions) is nearly equal on both sides of the wall and does not drive net flow; oncotic pressure specifically refers to the osmotic effect of plasma proteins, which are unevenly distributed. |
| "Edema always means too much blood volume." | Edema can arise from four independent mechanisms — high P_c, low π_p, high capillary permeability, or lymphatic obstruction — with blood volume often normal or even low (e.g. nephrotic syndrome). |
| "All capillaries are equally leaky." | Permeability (Kf, σ) varies enormously by tissue: continuous (muscle, brain) vs. fenestrated (kidney, gut) vs. sinusoidal (liver, spleen, marrow). |
| "The lymphatic system is a passive drain, like a sink." | Lymph flow is actively propelled by lymphatic smooth muscle contraction, valves, and skeletal muscle pumping — impairment (e.g. after lymph node removal) causes lymphedema even with normal Starling forces. |
The scientist behind it
Ernest Henry Starling was a British physiologist whose career, mostly at University College London, produced two of physiology's foundational principles: the law of the heart (with William Bayliss and later formalized as the Frank–Starling mechanism) and the theory of capillary fluid exchange. He and Bayliss also co-discovered secretin in 1902, the first hormone to be identified, and Starling coined the very word "hormone" in 1905. His experimental style was characteristically direct: rather than theorize about tissue fluid balance in the abstract, he cannulated lymphatics in perfused animal limbs and altered one variable at a time — venous pressure, then plasma protein concentration — watching lymph flow respond in real time. This insistence on manipulating a single physical variable and observing a quantitative, reversible physiological response became a template for cardiovascular and renal physiology throughout the 20th century.
Origins · 1 minErnest Henry Starling and how this was discovered
Where this matters today
Critical care and shock management. In septic shock and burns, capillary permeability rises dramatically (increased Kf and reduced σ), causing massive fluid and protein leak into tissues; understanding the Starling forces guides fluid resuscitation strategy, choice between crystalloids and colloids, and recognition of "third-spacing."
Nephrology. Nephrotic syndrome, characterized by heavy proteinuria and hypoalbuminemia, is directly explained by the oncotic term of the Starling equation, and monitoring serum albumin guides both diagnosis and edema management.
Cardiology. Right and left heart failure raise systemic or pulmonary capillary hydrostatic pressure respectively, producing peripheral edema or pulmonary edema; diuretics, venodilators, and positioning are chosen specifically to reduce P_c.
Oncology and surgery. Lymph node dissection (e.g., in breast cancer surgery) or radiation-induced lymphatic damage impairs lymphatic return, producing chronic lymphedema — a direct clinical demonstration of what happens when the "backup drainage system" for capillary filtrate fails, independent of any change in Starling forces themselves.
Pharmacology. IV albumin, hydroxyethyl starch, and other colloid solutions are designed specifically to raise plasma oncotic pressure and pull fluid back into the vasculature — direct clinical exploitation of the π_p term.
High-altitude and aerospace medicine. Pulmonary and cerebral edema at high altitude, and fluid shifts in microgravity, are both analyzed using Starling-force reasoning applied to specialized capillary beds, connecting directly to gas exchange physiology in Gas Exchange & Diffusion.
Agriculture and veterinary medicine. Edema disease in livestock and management of intravenous fluid therapy in large animals both rely on the same quantitative framework, scaled to different baseline pressures and protein concentrations.
:::quiz[In the Starling equation Jv = Kf[(Pc − Pi) − σ(πp − πi)], what does σ (the reflection coefficient) represent?]
- The rate of blood flow through the capillary
- How effectively the capillary wall excludes plasma protein from crossing ✓
- The surface area of the capillary bed
- The concentration gradient for oxygen
Ask the tutor: How does the endothelial glycocalyx change the classical picture of arteriolar-end filtration versus venous-end reabsorption?
Why doesn't all the fluid that filters out of capillaries just stay in the tissues?
Most of it is returned by the lymphatic system, not by capillary reabsorption alone. Classical teaching described filtration at the arteriolar end and reabsorption at the venous end as roughly balanced, but more precise microperfusion studies show that filtration actually continues along most of the capillary's length, with the modest 2–4 L/day surplus captured by blind-ended lymphatic capillaries. These vessels use overlapping endothelial flaps as one-way valves, and lymph is propelled toward the subclavian veins by smooth muscle contraction, one-way valves, and surrounding skeletal muscle activity. If lymphatic drainage is impaired — for example, after lymph node removal in cancer surgery — edema results even though the Starling forces themselves are unchanged.
What is the difference between oncotic pressure and osmotic pressure?
Total osmotic pressure reflects all dissolved solutes, dominated by small ions like Na⁺ and Cl⁻, which are present at nearly equal concentrations on both sides of the capillary wall — so they generate no net osmotic gradient favoring fluid movement in either direction. Oncotic (colloid osmotic) pressure refers specifically to the osmotic effect of large plasma proteins, chiefly albumin, which cannot cross an intact capillary wall in significant quantities and are therefore unevenly distributed. This protein-specific imbalance is what actually drives net fluid movement in the Starling equation; it is a small fraction of total osmotic pressure but the only fraction that matters for capillary fluid balance.
Why do fenestrated and sinusoidal capillaries exist if continuous capillaries protect tissues from unwanted protein leakage?
Different organs need different exchange capacities. Continuous capillaries in muscle, skin, and brain restrict solute and fluid movement to protect delicate tissue environments (the blood–brain barrier is an extreme continuous-capillary specialization). Fenestrated capillaries in the kidney glomerulus and intestinal villi allow much higher fluid and small-solute flux, essential for filtration and nutrient absorption, respectively. Sinusoidal capillaries in the liver, spleen, and bone marrow have large gaps that let even plasma proteins and blood cells pass, which is necessary for the liver to secrete newly synthesized proteins directly into blood and for the spleen and marrow to exchange whole cells. Capillary structure is matched to each organ's physiological task.
How does IV albumin infusion help treat edema, and when does it not help?
Infusing concentrated albumin raises plasma oncotic pressure (π_p), directly strengthening the inward-pulling Starling force and helping draw fluid from the interstitium back into the vasculature. This is useful when the underlying problem is a genuine oncotic deficit, as in severe hypoalbuminemia from liver failure or nephrotic syndrome. It is far less effective, and sometimes counterproductive, when edema is driven by elevated capillary hydrostatic pressure (as in heart failure) or increased capillary permeability (as in sepsis), because in those cases the infused albumin may itself leak into the interstitium through the abnormally permeable wall, worsening interstitial oncotic pressure and edema rather than resolving it.
What happens to capillary exchange during a burn injury?
Burns and other severe trauma dramatically increase capillary permeability, raising both the filtration coefficient (Kf) and lowering the reflection coefficient (σ) for protein, effectively opening the normally tight junctions and glycocalyx barrier. Plasma proteins now leak freely into the interstitium, collapsing the normal oncotic gradient that opposes filtration and causing massive fluid shifts from blood into tissue — this is the physiological basis of burn shock, with dramatic edema at the burn site and, in large burns, systemic hypovolemia severe enough to require aggressive IV fluid resuscitation calculated by formulas such as the Parkland formula.
Is capillary exchange the same mechanism as gas exchange in the lungs?
They occur across structurally similar thin capillary walls but rely on different physical mechanisms for different purposes. Capillary fluid exchange, as described by the Starling equation, is dominated by bulk flow of water and small solutes driven by hydrostatic and oncotic pressure differences. Gas exchange for O₂ and CO₂ occurs almost entirely by simple diffusion down partial pressure gradients through the lipid membranes of the endothelium and alveolar epithelium, essentially independent of net fluid movement. Both processes happen simultaneously across the same capillary walls, but understanding one does not require assuming the other — gas exchange is treated in depth separately.
Why can standing still for a long time cause ankle swelling in a healthy person?
Prolonged standing increases hydrostatic pressure in leg veins and capillaries due to the added weight of a tall column of blood (gravitational effect), transiently raising P_c in the lower limbs well above its usual seated or supine value. This shifts the Starling balance toward increased net filtration in the legs and feet, and because lymphatic return and the skeletal muscle pump (normally activated by walking) are underused when standing motionless, filtered fluid accumulates faster than it is cleared, producing mild dependent edema. This is normal and reversible physiology, distinct from pathological edema, and resolves with walking, leg elevation, or compression stockings that mechanically raise interstitial pressure and support venous return.