Many patients experience recurrent ischemic events despite optimal antiplatelet therapy. This has generated much interest in finding a laboratory test of platelet function to identify such patients, who have been termed 'nonresponders' or antiplatelet 'resistant'. Laboratory tests of platelet function have identified 'resistance' in 5-60% of patients taking aspirin and 4-30% of those taking clopidogrel. However, these tests of 'resistance' have not correlated closely with subsequent recurrent events, and have not reliably identified nonresponders to antiplatelet therapy. Here, we identify and discuss three major limitations common to all these tests. Firstly, they are performed on citrate-anticoagulated blood, secondly, blood is stored for a variable period of time, and thirdly, the assessment of thrombotic status on the basis of platelet response to only one or two agonists ignores the complexity of the mechanism of platelet thrombus formation in vivo. In this Review we discuss the significance of these important limitations, and the applicability of such in vitro platelet function tests to the prediction of in vivo events. We conclude that such tests are so unphysiological that they cannot reliably predict the true thrombotic status of patients. Identification of 'resistance' on the basis of these tests lacks sensitivity and specificity for identifying thrombotic risk, and is likely to be artifactual.