The relation of the risk of venous thromboembolism to the use of oral contraceptives was assessed in a hospital-based study of 61 women suffering from a first episode of idiopathic deep vein thrombosis or pulmonary embolism (cases) and 1278 women admitted for trauma or respiratory infections (controls). Twenty (33%) of the cases and 121 (9%) of the controls had used oral contraceptives within the previous month, yielding an age-adjusted relative risk estimate of 8.1 (95% confidence interval 3.7 to 18) for recent users relative to never-users. For women using oral contraceptives containing less than 50 micrograms estrogen, the relative risk estimate was 11 (3.7 to 22); for preparations with 50 micrograms estrogen, it was 5.5 (2.1 to 15); and for preparations with more than 50 micrograms estrogen, it was 11 (3.9 to 30). Past use of oral contraceptives was not associated with an increased risk. The data suggest that the risk of venous thromboembolism is increased for recent oral contraceptive users relative to nonusers, even if women use oral contraceptives containing low doses of estrogen. Confidence intervals were wide, however, so that a reduction in the risk for users of lower dose formulations relative to users of higher dose formulations cannot be ruled out. Selection bias, if present, would have resulted in overestimation of the relative risk, but should not have distorted the comparisons according to dosage.