The habitat use is an individual choice that is influenced by physical conditions such as substrate type, food resources availability and adequate depth. However, habitat use is often measured only through interspecific variability because intraspecific variability is supposed to be low. Here, the differences in habitat use by two stream fish assemblages in two different environments (Brazilian rainforest and semiarid) were investigated at both interspecific and intraspecific levels. We performed 55 hours of underwater observation in a 200 meters long stretch in each stream and quantified the following habitat descriptors: (i) water velocity, (ii) distance from the stream bank, (iii) substratum, (iv) water column depth, (v) aquatic cover, and (vi) canopy percentage. To compare intra and interspecific variability we summarized the multivariate habitat use databases using Principal Components Analysis (PCA) on Euclidean distance. An Analysis of Similarity (ANOSIM) was performed to test the differences in habitat use by the two assemblages. Besides, in each fish community we did an Analysis of Variance (ANOVA) to test within vs between species variability for individual position on each PCA axes. To go further than these univariate tests, the differences among the species and assemblages were tested with Permutational Multivariate Analysis of Variance (PERMANOVA). The habitat use between assemblages was significantly different (ANOSIM – R=0.14; p<0.001). PERMANOVA revealed significant differences among species in both assemblages (Rainforest - F=7.25; p<0.001; semiarid - F=4.84; p<0.001). Lower F values in the semiarid assemblage revealed a higher level of intraspecific variability for this assemblage. Our findings showed high intra and interspecific variability in both stream fish assemblages and highlight the importance of measuring individual’s differences for this feature of fish biodiversity. Additionally, the versatility described for tropical freshwater fishes confirms the importance of measuring intraspecific variability from different ecological traits.