A long-standing debate concerns whether nectar sugar composition evolves as an adaptation to pollinator dietary requirements or whether it is 'phylogenetically constrained'. Here, we use a modelling approach to evaluate the hypothesis that nectar sucrose proportion (NSP) is an adaptation to pollinators. We analyse similar to 2100 species of asterids, spanning several plant families and pollinator groups (PGs), and show that the hypothesis of adaptation cannot be rejected: NSP evolves towards two optimal values, high NSP for specialist-pollinated and low NSP for generalist-pollinated plants. However, the inferred adaptive process is weak, suggesting that adaptation to PG only provides a partial explanation for how nectar evolves. Additional factors are therefore needed to fully explain nectar evolution, and we suggest that future studies might incorporate floral shape and size and the abiotic environment into the analytical framework. Further, we show that NSP and PG evolution are correlated - in a manner dictated by pollinator behaviour. This contrasts with the view that a plant necessarily has to adapt its nectar composition to ensure pollination but rather suggests that pollinators adapt their foraging behaviour or dietary requirements to the nectar sugar composition presented by the plants. Finally, we document unexpectedly sucrose-poor nectar in some specialized nectarivorous bird-pollinated plants from the Old World, which might represent an overlooked form of pollinator deception. Thus, our broad study provides several new insights into how nectar evolves and we conclude by discussing why maintaining the conceptual dichotomy between adaptation and constraint might be unhelpful for advancing this field.