The objective of this study was to evaluate the effects of harvesting methods on the macrofauna, a known indicator of soil quality that detects changes in the system and indicates management alternatives. The experimental design was randomized blocks with six replications, with each block consisting of four parcels that corresponded to treatments green cane (CC), burnt cane (CQ), green cane - burnt cane (CC-Q) and burnt cane - green cane (CQ-C). Samples were collected in February and July 2010. The animals were divided into major taxonomic groups and accounted. Were evaluated ecological indexes (Shannon, Pielou and richness) and average total density of individuals and groups. Data were analyzed by the nonparametric statistical tools by Friedman or Signal test at 5%. To check relationships between usoil fauna and environmental variables, was used a multivariate conditional ordination method, the redundancy analysis (RDA). The index richness was more efficient than the total average density to evaluate the influence of cane harvesting systems, with the highest values related to areas harvested without burning. Further, the occurrence of key groups in the areas harvested without burning configures the establishment of a trophic web. There was dominance of the social group Formicidae in all treatments.