The hypothesis that herbivory is higher in areas without fishing and will increase the rate at which hard coral communities return to pre-disturbance conditions was tested in and out of the marine protected areas (MPA) of Kenya after the 1998 El Niño Southern Oscillation (ENSO). Herbivory was estimated by assay and biomass methods, and both methods indicated higher herbivory in fishery closures. Despite higher herbivory, the effect of the ENSO disturbance was larger within these closures, with reefs undergoing a temporary transition from dominance by hard and soft coral to a temporary dominance of turf and erect algae that ended in the dominance of calcifying algae, massive Porites, Pocillopora and a few faviids six years after the disturbance. The fished reefs changed the least but had a greater cover of turf and erect algae and sponge shortly after the disturbance. Higher herbivory in the fishery closures reduced the abundance and persistence of herbivore-susceptible erect algae and created space and appropriate substratum for recruiting corals. Nonetheless, other post-settlement processes may have had strong influences such that annual rates of coral recovery were low ( approximately 2%) and not different between the management regimes. Recovery, as defined as and measured by the return to pre-disturbance coral cover and the dominant taxa, was slower in fishery closures than unmanaged reefs.