A fundamental prediction of life-history theory is that current reproduction reduces chances of future reproductive activity and success. In plants, however, short-term manipulative experiments often fail to support this. One reason may be the within-species variation in reproductive frequency and success among habitats and years. In this study, Ranunculus acris in three subarctic habitat types (a meadow forest, a fen woodland, and an alpine meadow) was prevented from seed production by flower bud removal during four years. The objective was to test if this treatment increased future reproductive activity relative to naturally reproducing plants. In the fifth year, all plants were left intact to test if natural, as compared to previously prevented reproduction, entailed negative effects on future fruiting success, seed set, fecundity and/or seed quality. In all habitat types, there were costs of reproduction in terms of reduced future reproductive activity, as naturally reproducing plants produced sexual shoots less frequently than plants for which flower buds were removed. In terms of other traits, fecundity costs of reproduction were detected at two habitats. In the meadow forest, naturally reproducing plants had fewer seeds, due to fewer flowers, compared to previously bud-removed plants. At the alpine meadow, reproduction entailed reduced future seed nitrogen content, and thus, potentially lower future offspring quality. In contrast, at the fen woodland, seeds of naturally reproducing plants weighed more than those of previously bud-removed plants. No costs were found in terms of fruiting success (the proportion of all flowers that set at least one seed), ovules per flower or seed set at any habitat. These results show that costs of reproduction varied among habitat types and were detectable in some, but not all selected traits. We should therefore not expect reproductive costs to be equally expressed in all environments.