Venous thromboembolism (VTE) is rare in young women but is associated with the use of combined oral contraceptives (OCs). In 1995 and 1996, three studies showed a difference in the risk of VTE with third-generation OCs containing the progestogens, desogestrel or gestodene, compared with earlier formulations. However, the subsequent MediPlus study did not show any difference in the risk of VTE between users of third- and second-generation OCs. To re-examine the risks of VTE with various OCs, a nested case-control study was undertaken using the General Practice Research Database (GPRD). This study identified 293 cases and selected up to four controls matched for year of birth, practice, and event date. Adjustment for confounding variables included: body mass index, smoking, asthma, diastolic blood pressure, and a proxy for recent illness. The new analysis of the GPRD showed that there was no statistically significant difference in the risk of VTE among users of third-generation OCs compared with second-generation OCs containing levonorgestrel 150 microg plus ethinylestradiol 30 microg. Important associations with idiopathic VTE included: age, obesity, smoking, recent concurrent illness, and asthma. Thus, any difference previously noted between OCs containing desogestrel or gestodene and levonorgestrel are likely to be due to the healthy-user effect, prescribing bias and inadequate control of known confounding variables, such as age and obesity.