MicroRNomics is a novel genomics that studies the identification, targets, biological functions, etc., of microRNAs (miRNAs) on a genomic scale. Computational target prediction algorithms are important applications in microRNomics. However, the overlaps between target sets predicted by different algorithms for one miRNA are often small. Our work is initiated to find the reasons causing "heterogeneity" and investigate whether the heterogeneous targets are homogeneous on functional levels by integrating similarity metrics. The results suggest that most human miRNAs own heterogeneous targets. The dissimilarity of thermodynamic characteristics and the different treatment of 3'-compensatory sites adopted by algorithms are the main reasons for target "heterogeneity." Meanwhile, we find most miRNA heterogeneous targets are functional homogeneity because of the common principles such as sites conservation and G:U wobble pairs in different algorithms. Our findings reveal the "functional homogeneity in miRNA target heterogeneity." The conclusions provide a perspective of microRNomics on functional levels, which introduce a new sight into human miRNA targets.