This study aimed to compare two plasma substitution regimens used during plasma exchanges (PE). It was a prospective cross-over randomized trial. Each patient (n = 12) had two PE at a 48 h interval. During one PE, only albumin was administered (PEA), and during the other one, equal volumes of albumin and low molecular weight hydroxyethylstarch (HES) (Elohes) were given (PEA+E). The order in which these different protocols were used was random. Plasma was separated by filtration, and the total volume extracted was one and a half the plasma volume. The parameters recorded every 15 min until 1 h after the end of PE, were heart rate, blood pressure and central venous pressure (CVP). Plasma volume, calculated from the mean body haematocrit and blood volume, was measured before and after PE. The clinical and biological tolerance of the rapid infusion of a large volume of HES was also assessed. PE characteristics were similar in both groups. For PEA and PEA+E, PE lasted 152 +/- 21 min and 154 +/- 25 min; the plasma volume extracted was 3,907 +/- 772 ml and 3,933 +/- 717 ml; the volume of plasma substitute infused was 4,097 +/- 617 ml and 3,933 +/- 717 ml, respectively. As haemodynamic and biochemical values were not significantly different in both groups, they were pooled together irrespective of the order of PE.(ABSTRACT TRUNCATED AT 250 WORDS)