We compared the safety, effect on plasma proteins, and contribution to hypocalcemic toxicity of 10% pentastarch vs. 5% albumin in plasma exchange. Thirty‐two neurology patients underwent 161 plasma exchange procedures. Subjects were randomized to receive either 10% pentastarch (n = 17) or 5% albumin (n = 15) as the first ½ of colloid replacement. The second ½ of colloid replacement was 5% albumin in all cases. NaCl (plus small amounts of ACD‐A) accounted for ¼ of the total return fluid. Mean total exchange volume was 3,842.6 ± 450 ml. Hemoglobin, platelet count, coagulation parameters, and plasma fibrinogen were similar between the two groups at the outset of plasma exchange and at the time of the last (4th or 5th) procedure of the series. Immunoglobulin levels were equivalent at the outset and were reduced by plasma exchange to the same extent in the two groups. Serum albumin concentration fell significantly faster in the group receiving pentastarch. On the other hand, serum calcium, corrected for the albumin concentration, fell significantly further in the control group than in the group receiving pentastarch. Mean serum ionized calcium fell ∼25 % in procedures using only albumin but only ∼16% when pentastarch was used (P < 0.0001). This was reflected in the occurrence of hypocalcemic toxicity in 33.3% of procedures performed using only albumin but in only 8.1% of those using pentastarch (P = 0.0002). Pentastarch may be suitable for partial colloid replacement in plasma exchange and would have saved our hospital $127,050 to $434,280 in 1998 if used in all procedures. J. Clin. Apheresis 14:114–121, 1999. © 1999 Wiley‐Liss, Inc.