Important criteria for assessing a cell separator are thrombocyte yield, separation efficiency and purity of the thrombocyte concentrates. Based on a Multicentric Counting Study in which twelve centers participated, we conclude that it is very difficult to compare the results of the various centers in regard to the separation efficiency. This is especially true for the comparison of different separation procedures. In Marburg we compared three different cell separators of the newest generation. COBE Spectra (n = 71), Fresenius AS-104 (n greater than 1100) and Fenwal CS-3000 TNX (n = 79). The COBE Spectra exhibited the best separation efficiency with the lowest leukocyte contamination (thrombocytes 4.3 x 10(11) (72.2%), leukocytes 0.5 x 10(7)) on the condition that the ACD-blood ratio did not differ more than -15% from the required algorithm. In order to reduce the risk to the donor, the system correspondingly reduces the donor's blood flow, resulting in a longer donation time (on the average 89-100 min). When the ACD ratio was reduced further, a considerable number of spontaneous and sometimes irreversible platelet aggregation occurred, increasing the risk of reduced platelet function and shortened survival. The AS-104 and the modified CS-3000 (TNX) had similar separation efficiencies (approx. 60%). While the platelet concentrates (PC) of the AS-104 almost reached the purity of that from the COBE Spectra, the leukocyte contamination of the CS-3000 PC's was still about four times higher. Other results published show that morphology, in vitro function and in vivo survival of thrombocytes collected with the AS-104 are significantly better than those with the CS-3000. Based on further criteria, the three systems were compared with each other. However, we could not come to a definite recommendation for any of the systems.