OBJECTIVE: To investigate whether patients changing from their current oral contraceptive (OC) are preferentially switched to third-generation preparations. METHOD: The database comprised 12 months of medical records generated during 1990-91 from 693,705 patients and 153,687 OC users aged 14-45 years registered with 398 practices in the United Kingdom. Subjects who switched OC preparations were identified. RESULTS: A total of 16,197 women (10.5%) switched OC preparations at least once during the study period. There was a statistically significant net switching from second-generation OCs to third generation OCs with a ratio of 1.19 (95% CI, 1.12-1.26). This switching behavior took place in both younger (< 25 years) and older (> or = 25 years) OC users. In addition, a significant net switching was observed from progestogen-only pills to both second- and third-generation OCs. This was more pronounced in women > or = 25 years. CONCLUSION: Bias resulting from preferential switching of women from second-generation to third-generation OCs may have influenced the results of recently published studies on combined OCs and the risk of venous thromboembolism. Studies of venous thromboembolism in women using OCs should either restrict from the study women who have switched preparations or should include switching as a variable in any analysis.