The stressed bleeding time is a simple 'global' test of haemostasis, dependent upon platelet function, rheology, thrombosis and intimal function. It could be of considerable value in clinical practice if it were shown to be predictive of vascular disease events. A stressed bleeding time test was done on 1319 men aged 55-69 years in the Caerphilly Cohort Study of Heart Disease, Stroke and Cognitive Decline. The men were followed-up and during the following 7-10 years 155 men had a myocardial infarction (MI) and 72 an ischaemic stroke. The mean bleeding time was 323 (SD 113)s. This was shorter in men who smoked by an average of 45 s, and lengthened in men who took aspirin daily by 40s. After making statistical adjustments for numerous possible confounding factors, the relative odds (ROs) of an MI within the third of men with the longest bleeding times, compared to the third with the shortest times, was 0.90 (0.40-2.03). For ischaemic stroke, the ROs in the third of men with the longest times were 1.42 (0.39-5.21). The stressed bleeding time does not predict either MI or ischaemic stroke. It has no place in health screening.