Next-generation sequencing technologies have become a powerful tool for the analysis of microbial communities. Sequencing of the hypervariable regions of the 16S ribosomal RNA gene following the amplitag amplification process has allowed the study of the diversity of samples of diverse origin. According to previous reports, the number of sequences required for the correct determination of the composition of a given sample may vary with the degree of diversity of that sample. In this paper, we investigate the correctness of comparing heterogeneous size datasets of bird intestinal microbial communities obtained from pyrosequencing data (Roche 454 technology) without prior normalization. We conclude that the differences observed between samples are due mainly to individual differences, not to differences in the number of readings in each sample, which makes data normalization unnecessary with the conditions described here.