Does contraception contribute to better maternal health beyond its potential to reduce the proportion of births that are unwanted? The answer depends on the extent to which births averted by contraception would otherwise pose a greater risk to the mother than wanted or intended births. The risks of childbirth are known to vary with the mother's age and may also be linked to her parity and to the interval since the previous birth. Contraception is likely to change the age pattern of childbearing, particularly by reducing fertility at older ages, and will certainly affect parity-specific fertility. The first section of this paper reviews the evidence for the assertion that contraception benefits maternal health by reducing the number of high-risk births. The second section examines a related possibility - that unwanted births represent a greater threat to the mother's health than wanted births because less time and money are invested in antenatal and natal care. Lack of antenatal care and, to a greater extent, lack of skilled medical supervision during delivery are established risk factors for poor obstetric outcomes. This section presents new evidence on the link between unwantedness and obstetric care. The second major contribution of contraception to reducing obstetric mortality and morbidity is related to its potential to diminish recourse to unsafe abortion. Globally, it is estimated that 22% of pregnancies are aborted (Alan Guttmacher Institute, 1999). By definition these pregnancies are unwelcome, in the sense of being either absolutely unwanted or mistimed (i.e. the mother may want to have a child at some future date but not at the time of conception). In most industrialized countries, abortion is legally permitted for a wide variety of reasons, is performed by properly trained staff and carries very little risk to the physical well-being of the mother. Conversely, in most African and Latin American countries and in many Asian countries, abortion is legally permitted only in extreme circumstances and the vast majority of abortions performed are illegal. It has been estimated that in 1995 20 million illegal abortions were carried out, of which 19 million were in developing countries (Henshaw et al., 1999). Abortion legality and safety are strongly correlated. When a pregnancy is terminated by someone lacking the necessary skills or in an environment that does not conform to minimum standards of hygiene, the risk of serious complication to the woman is high. Globally, it is estimated that about 13% of pregnancy-related and birth-related deaths in women are caused by unsafe abortion. While it might seem self-evident that greater use of contraception will lead to a reduction in abortions - both safe and unsafe - this link has been challenged and, indeed, it is true that rising levels of contraceptive use can be accompanied by a rising incidence of abortion. The third section of this paper uses published data to re-examine the relationship between changes in contraceptive practice and abortion rates.