Using data from the 1983 Somali Family Health Survey, this study examines the polygamy-fertility relationship in a sample of 2662 currently married women aged 15-49. The 5 cities covered, the largest in the country, are Hargeisa, Kismayo, Burco, Mogadishu, and Baydhaba. Somali urban wives in polygamous unions differed in several ways from their counterparts in monogamous marriages, but the differences usually varied by age of wife. For example, 1) those in polygamous unions had much earlier ages at 1st marital cohabitation, except those aged 25-34; 2) women aged 25-34 were much more likely to be employed, have husbands present, and not to have used contraception if they were married to polygamists; and 3) the percentage of wives married more than once was far greater for wives of polygamists. The results of a multiple regression of fertility show that 1) the age at 1st marital cohabitation had a highly significant inverse relationship with the number of children ever born per wife; 2) employed wives aged 15-34 had a higher mean cumulative fertility than those not employed; and 3) at ages 35-49, employed wives had borne fewer children apiece than had other wives of similar age. An important finding is that for wives in the middle and late ages of childbearing, the association between polygamy and cumulative fertility was conditioned by whether or not they had been married more than once. Remarried women in polygamous unions had a lower than average fertility. Remarried mothers in polygamous unions had the lowest fertility of all; therefore, involuntary sterility does not explain the phenomenon. The findings support the interpretation that the time spent between sexual unions accounts for the lower fertility of remarried wives. Thus, this investigation does not support the contention that polygamy leads to high cumulative fertility rates in urban Somalia. Rather, the data show that polygamy and monogamy select women with different social characteristics, which are associated with different rates of cumulative fertility.