The ability to connect numbers and magnitudes is an important prerequisite for math learning, here referred to as number-magnitude skills. It has been proposed that working memory plays an important role in constructing these connections. The aim of the current study was to examine if working memory accounts for constructing these connections by testing whether development of number-magnitude skills can be explained by different components of working memory. Number-magnitude skills and working memory skills of 69 children were assessed at age 5, 5.5 and 6. Different results were found for different components of working memory. Whereas no effects were found for the visuospatial sketchpad and phonological loop in development of number-magnitude skills, the central executive predicted variance in intercept and slope of number-magnitude skills. The results of this study provide longitudinal evidence for the involvement of general cognitive skills (i.e. working memory) in the development of number-magnitudes connections. (author abstract)