A multinational hospital-based case-control study of the risk of venous thromboembolic disease associated with combined oral contraceptives (OCs) done in 1989-93 prompted a separate inquiry comparing the risk of venous thromboembolism (VTE) associated with low oestrogen (< 35 micrograms ethinyloestradiol) OCs containing levonorgestrel with risks in low oestrogen preparations containing the third-generation progestagens desogestrel or gestodene. This analysis of data from 9 countries, involved 769 cases and 1979 age matched hospital controls and, in one centre, 246 community controls matched on age and general practice. 137 cases and 203 controls were current users of levonorgestrel (odds ratio [OR with 95% confidence interval] 3.5 [2.6-4.7]), with non-users as the reference; 35 cases and 28 controls were current users of desogestrel (9.1 [4.9-17.0]), and 36 cases and 28 controls were current users of gestodene (9.1 [4.9-16.7]). The ratios of these risks, compared with levonorgestrel, were 2.6 (1.4-4.8) for both products separately. Risk estimates adjusted for body mass index (BMI) were 3.4, 7.3, and 10.2 for levonorgestrel, desogestrel, and gestodene, respectively, compared with non-users, and 2.2 and 3.0 for desogestrel and gestodene, respectively, compared with levonorgestrel. 48 (68%) cases and 48 (86%) controls exposed to desogestrel or gestodene were from the UK (Oxford region). In this centre risk estimates compared with non-users, adjusted for BMI, were 2.6, 5.3, and 5.7 for levonorgestrel, desogestrel, and gestodene, respectively. Current users of low oestrogen dose combined OCs containing desogestrel or gestodene appear to be at higher risk of VTE than users of combined OCs containing levonorgestrel. The possibility that these unexpected results on a secondary study objective are due to chance, bias, or residual confounding cannot be excluded entirely and the results need to be confirmed by independent studies. They are at variance with the apparently more favourable metabolic effects of the newer progestagens. Whether the new progestagens are associated with lower risk of arterial disease (stroke and myocardial infarction) must be evaluated further.