A knowledge of the limitations of automated platelet counting is essential for the effective care of thrombocytopenic patients and management of platelet stocks for transfusion. For this study, 29 external quality assessment specimen pools with platelet counts between 5 and 64 × 109/L were distributed to more than 1,100 users of 23 different hematology analyzer models. The same specimen pools were analyzed by the international reference method (IRM) for platelet counting at 3 reference centers. The IRM values were on average lower than the all-methods median values returned by the automated analyzers. The majority (∼67%) of the automated analyzer results overestimated the platelet count compared with the IRM, with significant differences in 16.5% of cases. Performance differed between analyzer models. The observed differences may depend in part on the nature of the survey material and analyzer technology, but the findings have implications for the interpretation of platelet counts at levels of clinical decision making.