A recirculating in vitro CAVH system was designed which generated pulsatile blood and filtrate flows. Monitors recorded hydrostatic pressures simultaneously in the arterial, venous and filtrate lines during varying plasma or blood flow rates and predilution (vs postdilution) replacement fluid flow rates. Similar hydrostatic pressure monitoring was carried out during multiple maneuvers to generate suction on the filtrate side of the hemofilter (Amicon D-20's and Renaflo's). With a plasma flow (Qp) of 100 cc/min and predilution replacement fluid infusion rate of 500 cc/hr, the arterial pressure was 5% greater than during postdilution (p less than 0.05). With a blood flow (Qb) of 50 cc/min, predilution fluid replacement rates of 500 and 1000 cc/hr, and vacuum suction applied to the filtrate compartment, the arterial pressure was 33% lower than during postdilution fluid replacement (p less than 0.03). Nonetheless, the ultrafiltration rate (UFR) was 10 to 30% higher (p less than 0.03). At many other combinations of Qp, Qb and replacement rates and modes, there were no significant changes in arterial pressure. Despite these arterial pressure changes, greater than 70% of the transmembrane hydrostatic pressure (TMP) was due to the negative pressure induced by filtrate suction (gravity, Gomco, wall suction, IMED). The actual pressure in the filtrate compartment measured during Gomco or wall suction was 3/4 of that stated by their gauges, presumably due to leakage. Maximum wall suction never generated TMP's greater than 150 mmHg.(ABSTRACT TRUNCATED AT 250 WORDS)