A model is proposed for decomposition of changes in total fertility by several of its proximate determinants. The new method has been derived from previous work in modelling marital fertility within the framework of generalized linear models. The discussion is illustrated by data from countries in Latin America and the Caribbean in which there has been at least one survey from the World Fertility Survey and from the Demographic and Health Surveys projects. The results of the new approach are then compared with those obtained from the application of Bongaarts's model of proximate determinants to the same data. The most important lesson is that, because of the intricacies of human reproduction, the models proposed to assess the importance of the most relevant inhibitory factors on the potential fertility of populations generate results which are subject to considerable variability, and can only be regarded as crude approximations of the real effects.