It has been shown by several workers that the passivation of GaAs surfaces using sulfides results in a large reduction in the surface recombination velocity accompanied by an increase in the band bending onn‐type samples. This apparently contradictory pair of results leads to the suggestion that the responsible electronic states are a midgap donor compensated by an acceptor near the valence‐band maximum. We explore the consequences of such a model, particularly when the midgap state is assumed to be a double donor. In the double donor case, simple qualitative arguments indicate that the surface recombination velocity can be reduced by a factor much greater than the reduction in surface‐state density. The model is consistent with observations made using a variety of experimental techniques. A correlation between the electronic states and surface chemistry is made, and the As and Ga antisite defects are discussed as candidates for the donor and acceptor states.