Hormone replacement therapy (HRT) and the combined oral contraceptive pill (COCP) are both commonly used for oestrogen replacement in premature ovarian failure but there is a lack of evidence of their effects. We compared the effects of combined HRT (Nuvelle) and the COCP (Microgynon 30) in 30 women with premature ovarian failure in a two year open-label randomised trial. 29 women who declined to take treatment were also followed using the same protocol. 36 women (61%) completed the trial (15 in the no treatment group; 12 in the HRT group; 9 in the COCP group). In comparison with the COCP, treatment with HRT increased bone mineral density at the lumbar spine at 2 years, which was the primary outcome measure (+0.038 g/cm2; 95% CI 0.002 to 0.073; p 0.040; linear regression using adjustment for baseline values). Bone turnover markers (P1NP and CTX) showed similar reductions in the two treatment groups. There were trends in favour of HRT in lipid profile, high sensitivity C-reactive protein, blood pressure and sexual function. The HRT group had a significantly greater reduction in menopausal symptoms at 24 months. Improvements in most symptom scores took longer in the COCP group. In the no treatment group, bone density dropped at all sites over the course of the trial. The no treatment group also performed poorly in comparison with the treatment groups in bone turnover, depression score and menopausal symptoms score. These findings will have important implications for counselling young women with premature ovarian failure on their choice of oestrogen replacement. We have shown that in many respects HRT performed superiorly to the COCP. However, further research is required to confirm these effects. The results from the no treatment group will enable women who choose to decline treatment to be counselled on the effects of this choice of management.