We consider a non-linear version of the Generalized Assignment Problem, a well-known strongly N P-hard combinatorial optimization problem. We assume that the variables are continuous and that objective function and constraints are defined by non-linear functions of the variables. A mathematical model is introduced and used to derive upper bounds on the optimal solution value. We present constructive heuristics, obtained from decomposition and non-linear programming tools, and a binary linear programming model that provides approximate solutions. By combining the various methods and a local search framework, we finally obtain a hybrid heuristic approach. Extensive computational experiments show that the proposed methods outperform the direct application of non-linear solvers and provide high quality solutions in a reasonable amount of time.