Three decades after its population collapse, the sea urchin Diadema antillarum has yet to return to pre-mortality densities in most Caribbean localities. Circumstantial evidence suggests high post-settlement mortality due to predation as a potential mechanism to explain the lack of, or the slow recovery of this sea urchin. In this study we experimentally reduced the densities of two reef fish, Thalassoma bifasciatum and Halichoeres bivittatus, at one site for one year and measured the effect of their removal in the abundances and size structures of D. antillarum in relation to that of a comparable control site. We hypothesized a significant increase in urchin abundance and a reduction in mean test diameter at the experimental site with respect to the control site after the reduction in fish abundance. Overall, the density of D. antillarum did not vary significantly between control and experimental sites for the different census periods. Fish removal had a significant effect on the size structure of D. antillarum and in the mean density of the medium size classes. Results suggest that removal of these two labrids can significantly increase the relative abundance of medium-sized sea urchins and modify its population structure. Efforts to recover D. antillarum populations would be strengthened if densities of T. bifasciatum and H. bivittatus remain low.