INDUCED ABORTION is associated with several clinical and demographic measures. Recent reports have demonstrated the relative safety of early surgical termination of -pregnancy (1-3) as well as the relationship between legal abortion and a decline in birth rates and fertility indices (4,5). New York City data suggest that out-of-wedlock births may also be decreased, and Evrard (6) has further suggested that widespread use of abortion may affect the maternal and perinatal mortality rates. In this paper we will examine the effect of varying utilization of induced abortion on the maternal mortality rate and the relationship between nonviable conceptions and the measurement of maternal risk. We will attempt to show the limitations of the traditional maternal mortality rate as a measure for describing or comparing maternal risks for different cohorts. Alternative methods of measurement will be discussed. The traditional maternal mortality rate is expressed as the ratio of the number of maternal deaths (numerator) to the live births (denominator) in a defined population per unit of time. It seems likely that the rationale for the traditional measure stems from the fact that live births were the most readily obtainable quantity useful for measuring maternal risk. In addition, we might speculate that it is derived from a notion of a cost to benefit ratio-the cost in maternal deaths for the benefit of a given number of live births, that is, the cost of successful reproduction. When the large majority of known pregnancies ended in live births, the traditional maternal mortality rate (TMMR) represented an adequate measure of pregnancy risk which was useful for both descriptive and comparative purposes. Because conceptions ending in spontaneous or induced abortion were difficult to ascertain, their relationship to the maternal mortality rate could not be fully appreciated. The legalization of induced abortion has resulted in a substantial increase in knowledge concerning the incidence of