OBJECTIVE--To examine the associations between air temperature and risk factors for ischaemic heart disease. METHOD--Data on risk factors are available from up to 2036 men in the Caerphilly Prospective Heart Disease Study. Daily temperatures were obtained from the Meteorological Office. Relations between these were examined by regression. RESULTS--The coldest month of the year has a mean temperature that is 16 degrees C lower than that in the warmest month. A fall in temperature of this magnitude is associated with higher blood pressures (by 3-5 mm Hg) and a lower concentration of high density lipoprotein cholesterol (by 0.08 mmol/l). The most important effects however, seem to be on the haemostatic system. Fibrinogen is 0.34 g/l higher in the coldest month than in the warmest (p < 0.001) and alpha 2 macroglobulin, a protein that inhibits fibrinolysis, is also raised. Platelet count is increased by 30% of a standard deviation and the sensitivity of platelets in whole blood to adenosine diphosphate is increased by cold. CONCLUSIONS--These effects on haemostasis, together with the effect on blood pressure, could explain a large part of the increase in ischaemic heart disease in the winter but are unlikely to explain much of the difference in mortality within different areas of England and Wales.