The analysis of endogenous growth models with pollution often concentrates on steady state trajectories, under the assumption that the steady state is in some sense stable. In the present note we provide examples showing that this issue should be dealt with carefully. We use the Rebelo Ak model augmented with a stock of pollutants causing a negative externality. It is found that optimal growth is not necessarily balanced (contrary to the outcome of the standard Rebelo model). Moreover, the existence of the externality may affect long run optimal growth rates.