A large-scale longitudinal study was conducted in which the elementary mathematical knowledge and skills of a large group of Flemish preservice elementary school teachers from 15 different institutes was assessed by means of a paper-and-pencil test that was administered both at the beginning and at the end of their 3-year training. The 30-items test covered the new standards for mathematics in the elementary school curriculum in Flanders. The test was divided in six subsets differing in terms of the curricular subdomain and of the cognitive operations being addressed by the item. The results confirmed the frequently heard concern that at the beginning of their training preservice elementary school teachers have rather weak mathematical competencies. At the end of their 3-year training, the overall test performance had become substantially better, although there were still reasons to be seriously concerned about the readiness of some student teachers to teach mathematics to elementary school children. A number of more specific comparisons helped to identify the relative role of two different factors-selection and instruction-in student teachers' gain from pretest to posttest. Besides documenting the development of elementary mathematical competence in preservice elementary school teachers, the study also resulted in an instrument for (self-) assessment of elementary school mathematical competence. This instrument is now being used in many institutes for elementary school teacher training in Flanders.