The relative cost-effectiveness of different treatment strategies for hormone replacement therapy (HRT) was assessed within the framework of a computer model. Where data were lacking, it was necessary to make assumptions about the effects of HRT, particularly in relation to combined oestrogen-progestogen therapy and cardiovascular disease; however, sensitivity analyses were performed to assess the impact of changing these assumptions on the cost-effectiveness equation. It appears that net expenditure by the NHS will depend critically on the direct costs of treatment, rather than on any indirect costs incurred or saved as a result of side-effects. In terms of mortality, a reduction in cardiovascular disease risk would have greatest impact and would overshadow any small increase in breast cancer risk which may be associated with long-term use. If the cardioprotective effect of oestrogen is real, our results suggest that long-term prophylactic treatment of hysterectomised women would be relatively cost-effective. Treatment of symptomatic menopausal women for any period of time appears to offer very good value for money. The lack of data relating to combined oestrogen-progestogen therapy and cardioprotection, and the major importance of the latter in the equation of benefits and risks, make it more difficult to draw conclusions about the cost-effectiveness of treating non-hysterectomised asymptomatic women for prophylactic reasons.