OBJECTIVE--To study the outcome of pregnancy in women with artificial heart valves treated in major European centres, and to compare the safety and efficacy of different anticoagulant regimens and of mechanical and bioprosthetic valves. DESIGN--Retrospective study. METHOD--The information was obtained by questionnaire sent to all major cardiac centres in Europe sending one cardiologist from each centre a covering letter suggesting that the questionnaire be handed on where appropriate to the most relevant colleague. RESULTS--214 pregnancies were reported in 182 women; 151 pregnancies in 133 women with mechanical valves, and 63 pregnancies in 45 women with bioprostheses. Most women were in New York Heart Association class I or II and in sinus rhythm. 150 women with mechanical valves and 11 (17%) with bioprostheses received anticoagulants during pregnancy. One patient with an aortic valve prosthesis refused to take anticoagulants. Including the spontaneous abortions reported as well as the therapeutic abortions 83% of the pregnancies in women with bioprostheses and 73% in those with mechanical valves resulted in a healthy baby, full term or premature, who did well (NS). The incidence of stillbirths (3% and 6%) and of reported spontaneous abortions are excluded (because they are unlikely to have been fully reported), the success rate was 91% for women with bioprosthetic valves and 84% for women with mechanical valves (NS). 114 (53%) of the women had taken warfarin for some part of the pregnancy (46 (40%) of these during the first trimester) but there were no embryopathies. There were 13 valve thromboses (four fatal), eight embolic events (two fatal), and seven bleeds in women with mechanical valves. Most of these complications occurred with heparin but fatal aortic valve thrombosis occurred in the one woman who refused anticoagulant treatment out of the 151 women with mechanical valves. There were no maternal deaths in the bioprosthesis group but in 17/49 (35%) of these valves functional deterioration led to urgent replacement during pregnancy (two) or soon after. CONCLUSIONS--The outcome of pregnancy was similar for women with mechanical valves or bioprostheses. Warfarin treatment was safe and effective and was not associated with embryopathy. Heparin treatment was associated with more thromboembolic complications and more bleeding complications. Bioprostheses deteriorate rapidly during pregnancy.