Objective: To determine the true platelet count in blood salvaged and processed for autotransfusion. Design: Prospective, randomized study. Setting: Department of anesthesiology and orthopedic clinic of a university hospital. Patients: 60 patients who were scheduled for elective hip surgery and included in routine intra- and postoperative blood salvage and autotransfusion programs. Interventions: Patients were randomly divided into 2 groups. In group I (n = 30) autotransfusion was performed using the Continuous Autotransfusion System^® (Fresenius), in group II (n = 30) the Cell Saver 5^® (Haemonetics) was used. In both groups samples were taken from the collection reservoir and the washed red cells. Platelet counts were performed with an automated cell analyzer and by flow-cytometric analysis using the monoclonal antibody CD42a. Results: The platelet counts obtained by the automated analyzer were 5 to 21 times higher than those obtained by flow-cytometric analysis. This appears to be due to the contamination of the samples with cell debris of about the same size as platelets. The platelet counts stated in literature should be reconsidered. Conclusions: Automatic cell-counting devices are inadequate to determine the platelet count of blood salvaged and processed for autotransfusion.