We have prospectively studied and compared two consecutive groups of critically ill patients treated with either continuous arteriovenous hemodiafiltration (CAVHD) (n = 28) or continuous venovenous hemodiafiltration (CVVHD) (n = 25) to establish the technique of choice. The two groups were comparable in mean age (59 v 58 years), mean Acute Physiology and Chronic Health Evaluation (APACHE) II score (29.6 v 27.4, P = NS), requirements for inotropic drugs, and mean number of failing organs (2.9 v 3.2). CVVHD led to a greater amount of hourly ultrafiltrate (mean, 590 v 424 mL; P < 0.001), but urea and creatinine clearances were not significantly different with the two techniques. Twelve patients survived in the CAVHD group (42.8%) and 13 in the CVVHD group (52%; P = NS). The major advantage for CVVHD use was the substantial decrease in the number of access-related complications (2 v 10; P < 0.025). We conclude that while CVVHD does not offer a significant increase in solute clearance, it significantly minimizes vascular access-related morbidity and should therefore be regarded as the therapeutic modality of choice.