Apparent permeability to albumin (P(alb)) was measured with (125)I-albumin in specimens of rabbit parietal pericardium from lumen to interstitium (L-I) and from interstitium to lumen (I-L). With albumin concentration (C(alb)) 0.5%, P(alb) (x 10(-5) cm/s) L-I at 37 degrees C was 0.172 +/- 0.019 SE; it decreased to 0.092 +/- 0.022 I-L at 37 degrees C, 0.089 +/- 0.021 L-I at 12 degrees C, and 0.084 +/- 0.018 I-L at 12 degrees C. These findings provide evidence for an active transport L-I, likely transcytosis. With C(alb) 2.5%, 0.05%, and 0.005%, P(alb) L-I at 37 degrees C was 0.188 +/- 0.023, 0.156 +/- 0.021, and 0.090 +/- 0.021, respectively; at 12 degrees C it was 0.089 +/- 0.017, 0.083 +/- 0.019, and 0.087 +/- 0.026, respectively. Hence, active albumin transport ceases with C(alb) 0.005%; P(alb) values I-L at 12 degrees C and with C(alb) 0.005% are similar and provide diffusional permeability. With physiological C(alb) (approximately 1%), active albumin flux was approximately 5 x 10(-4) micromol x h(-1) x cm(-2). Apparent permeability to FITC-dextran 70 (P(dx)) was also measured. P(dx) (x 10(-5) cm/s) L-I at 37 degrees C with C(alb) 0.5% was 0.095 +/- 0.018; it decreased to 0.026 +/- 0.004 I-L (37 degrees C, C(alb) 0.5%), 0.038 +/- 0.007 at 12 degrees C (L-I, C(alb) 0.5%), 0.030 +/- 0.009 with C(alb) 0.005% (L-I, 37 degrees C), and 0.032 +/- 0.011 with nocodazole (L-I, 37 degrees C, C(alb) 0.5%). These findings provide evidence for transcytosis and confirm conclusions drawn from P(alb). Vesicular liquid flow, computed from vesicular dextran flux (fluid-phase only), was approximately 3.5 microl x h(-1) x cm(-2). Transcytosis seems a relevant mechanism, removing protein and liquid from serous cavities.