One of the most important questions in ecology is the relative importance of local conditions (niche processes) and dispersal ability (neutral processes) in driving metacommunity structure. Although many studies have been conducted in recent years, there is still much debate. We evaluated the processes (niche and neutral) responsible for variation in anuran composition in 28 lentic water bodies in southeastern Brazil. Because anurans depend heavily on environmental conditions, we hypothesized that environmental variables (niche processes) are the most important drivers of community composition. Additionally, as anurans have limited dispersal abilities, and the study region presents harsh conditions (high forest fragmentation, low rainfall and long dry season), we expected a lower, but significant, spatial signature in metacommunity structure, due to neutral dynamics. We used a partial redundancy analysis with variation partitioning to evaluate the relative influence of environmental and spatial variables as drivers of metacommunity structure. Additionally, we used a recently developed spatial autocorrelation analysis to test if neutral dynamics can be attributed to the pure spatial component. This analysis is based on predictions that species abundances are independent but similarly spatially structured, with correlograms similar in shape. Therefore, under neutral dynamics there is no expectation of a correlation between the pairwise distance of spatial correlograms and the pairwise correlation of species abundances predicted by the pure spatial component. We found that the environmental component explained 21.5%, the spatial component 10.2%, and the shared component 6.4% of the metacommunity structure. We found no correlation between correlograms and correlation of abundances predicted by the pure spatial component (Mantel test = -0.109, P = 0.961). In our study, niche-based processes are the dominant process that explained community composition. However, neutral processes are important because spatial variation can be attributed to pure neutral dynamics rather than to missing spatially structured environmental factors.Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)