The main goal of this study was to describe the spatial and temporal distribution of candidates for blood donation in Belo Horizonte, Minas Gerais State, Brazil, who appeared at the Hemominas Foundation in 1994 and 2004. The study also compared the candidates for age, gender, and clinical approval for blood donation in space and space-time. Data were obtained from a cross-sectional study for 1994 and were randomly selected from all donor candidates for 2004. The samples were georeferenced using the residential address. The spatial analysis techniques employed were: Kernel maps, thematic maps of Bayesian empirical rates and crude rates, and Moran Global. According to the findings, spatial distribution of candidates was non-random. The Kernel maps helped detect points with higher or lower concentration of candidates. Thematic maps described the concentration of candidates in relation to the population for the various categories. The results could help detect areas for actions targeting donor recruitment and areas with specific blood donation public campaign needs.