Wild bee communities were studied on one- to five-year-old set-aside fields with naturally developed vegetation (n=20), and old orchard meadows (n=4) to analyse effects of secondary succession on species diversity, resource use and associated life history traits. General theory predicts a steady increase of species richness with age of succession. In contrast, we found a first maximum in species richness of bees on two-year-old set-aside fields and a second on old meadows. Successional changes of bee communities were related to changes of vegetation. The transition from pioneer successional stages, dominated by annuals, to early successional stages, dominated by perennials, resulted in the highest species richness of flowering plants in the second year within the first five years of succession. Species richness of flowering plants was the best predictor variable for species richness of bees, whereas the cover of flowering plants correlated with the abundance of bees. Annual plants were visited more often and perennials less often than expected from their flower cover. Halictidae tended to prefer flowers of annuals, whereas Megachilidae, Apidae and Anthophoridae significantly preferred perennials. In departure from successional theory, body size, proportion of specialised bees and proportion of parasitic bees did not significantly increase with successional age, but number of generations and the proportion of soil-nesting bees decreased with successional age. Comparison of different management types showed that set-aside fields with naturally developed vegetation supported much more specialised and endangered bee species than set-aside fields sown with Phacelia tanacetifolia.