Abstract: Understanding the pattern in spatial distribution of soil microbial processes is critical to understand the environmental factors that regulate them as well as to scale up these processes to ecosystem. Soil samples from a 1. ha tropical rainforest plot (Paracou, French Guiana) were analyzed according a nested sampling approach using different separation distances ranging from 0.4 to 40. m. The variability of substrate induced respiration (SIR) and of denitrification enzyme activity (DEA) was characterized in relation to various soil properties (total C and N contents, NIRS related index of soil organic matter quality, SOMQ, and index of tree influence potential, IP). The variability of SIR and DEA was higher than that of environmental properties. The patterns of accumulated variance as a function of distance varied among the soil properties. The variability of SIR and DEA mainly occurred at small (1. m) scale (and at the 10-40. m-scales for SIR), probably reflecting the quality of litter input that results of the influence of local assemblage of different tree species, though changes in the soil N and C contents. Indeed, total soil C and N contents explained the microbial properties at every scale. Coefficients of codispersion showed that neither SOMQ nor IP did correlate with SIR and DEA, and confirmed that total C and N contents explained microbial properties in a scale dependent and complex manner. Such spatial dependency underlines the importance of soil heterogeneity in this tropical forest with implications for sampling strategies when studying the microbial processes and their response to disturbances. © 2012 Elsevier B.V.