Data from the 1990 Demographic and Health Survey (DHS), the 1985 census, and PROFAMILIA and Ministry of Health service statistics were used to analyze persistence of high fertility in different subregions and socioeconomic groups of Colombia. The 1990 DHS sample design divided the country into 13 subregions, allowing greater than usual disaggregation of data. The analytic strategy had three parts: identification of regions exhibiting high fertility during 1987-90; characterization of the regions according to macro level indicators and fertility level; and analysis of the importance of the effects of the contextual and individual variables on recent fertility by means of a multilevel multivariate model. The characterization of the regions and the multivariate analysis tested the hypothesis that contextual conditions influence fertility directly and not just as instruments of the individual characteristics of local populations. Based on demographic transition theory and available information, several contextual indicators were studied: women's status, economic role of children, infant mortality, access to modern family planning methods, and urbanization. The analysis demonstrated the existence of high fertility in 3 of the 13 geographic subregions: northeast, northwest, and Tolima Grande. The characterization of the subregions indicated that those where women had lower status, and where there was less emphasis on children's school attendance, high infant mortality, low access to family planning, and low level of urbanization were not necessarily the areas with the highest total fertility rates, suggesting that a cultural effect might also be present. The northeast and northwest subregions have cultural values and family structures different from those of the rest of Colombia and similar to other Caribbean countries: prevalence of consensual unions, early marriage, and high value of children in the household. The results of the multivariate analysis also indicated the presence of a distinct cultural effect on fertility.