Samples of venous and capillary blood were collected simultaneously from healthy adults to assess the accuracy of platelet counts in capillary blood as determined by an automated particle counter. The difference between the mean venous blood platelet count (248,300) and the mean capillary blood count (215,500) was highly significant (P < .001). For 24% (7/29) of the subjects, the capillary blood platelet count underestimated the venous blood count by ≥25%, with three subjects erroneously classified as thrombocytopenic. A heterogeneous group of thrombocytopenic patients showed a similar difference in mean platelet counts (venous blood: 72,500/μ1; capillary blood: 65,400/μ1; P = 0.01). In most clinical situations, capillary blood platelet counts were adequate for patient evaluation; however, when an accurate platelet count is necessary, venous blood should be used.