No-take marine reserves are one of the oldest and most versatile tools used across the Pacific for the conservation of reef resources, in particular for invertebrates traditionally targeted by local fishers. Assessing their actual efficiency is still a challenge in complex ecosystems such as coral reefs, where reserve effects are likely to be obscured by high levels of environmental variability. The goal of this study was to investigate the potential interference of small-scale habitat structure on the efficiency of reserves. The spatial distribution of widely harvested macroinvertebrates was surveyed in a large set of protected vs. unprotected stations from eleven reefs located in New Caledonia. Abundance, density and individual size data were collected along random, small-scale (20×1 m) transects. Fine habitat typology was derived with a quantitative photographic method using 17 local habitat variables. Marine reserves substantially augmented the local density, size structure and biomass of the target species. Density of Trochus niloticus and Tridacna maxima doubled globally inside the reserve network; average size was greater by 10 to 20% for T. niloticus. We demonstrated that the apparent success of protection could be obscured by marked variations in population structure occurring over short distances, resulting from small-scale heterogeneity in the reef habitat. The efficiency of reserves appeared to be modulated by the availability of suitable habitats at the decimetric scale ("microhabitats") for the considered sessile/low-mobile macroinvertebrate species. Incorporating microhabitat distribution could significantly enhance the efficiency of habitat surrogacy, a valuable approach in the case of conservation targets focusing on endangered or emblematic macroinvertebrate or relatively sedentary fish species.