Several previous studies have attempted to correlate habitat complexity and reef fish species diversity. These studies have mostly examined natural reef systems, but results differed. To examine this relation, we built 1 m 2 habitats with 20 replicates of five complexity levels from July to August 2001 in the northeastern Gulf of Mexico (n=100). In June and July 2002, we built new habitats using the 2001 design, but also added a sixth complexity level (n=120). In order of increasing complexity these included: cage, shell, cage-shell, block-shell, cage-block-shell, and shellblock-pyramid habitats. Most fish in both years were juveniles and included species common to reef structures in the northeastern Gulf of Mexico. In 2001, we identified 26 fish species, and the dominant species was red snapper, Lutjanus campechanus (41%), followed by rock sea bass, Centropristis philadelphica (23%), and sand perch, Diplectrum spp. (14%). In 2002 we identified 36 species, and the dominant species was tomtate, Haemulon aurolineatum (36%), followed by Diplectrum spp. (19%), and L. campechanus (13%). In 2001, species diversity and richness were significantly (P < 0.05) higher on more complex habitats (H¢=1.7, S=11-12) compared to less complex habitats (H¢=0.8-1.0, S=4-9). In 2002, patterns among diversity, richness and reef complexity were less apparent with only the least complex habitats shell and cage showing significantly lower values. In both years, multidimensional scaling grouped by complexity levels with cage and shell habitats showing the clearest separation from other habitat types. Also, with few exceptions (only 8%) analysis of similarities showed significant (P < 0.05) differences in fish communities across complexity levels. Although community composition varied between years, this study provided evidence to support the hypothesis that habitat complexity increased reef fish species diversity.