MicroRNAs (miRNAs) are known to play important roles in the way their target genes behave. By causing translational repression or degradation of their targets, miRNAs affect many biological functions and in turn are linked to many diseases. One way of understanding the functions of miRNAs is by identifying and analyzing their target genes through lab experiments. However, due to the inefficiency of this method, alternative ways had to be explored, one of which is the use of computational and bioinformatics approaches. Even though these approaches can save a lot of experimental time and chemical cost, their results can be questionable. In this study, we evaluated three commonly used target prediction tools by analyzing their predicted targets through miRNA-mRNA network. Our approach combines the use of gene function similarity and gene expression data in determining the most probable true targets. Our results show wide outspread of predicted targets with less overlap among the tools. Overall, TargetScan seems to perform better compared to the other two.