MicroRNAs (miRNAs) are a class of non-protein coding small RNAs that regulate gene post-transcriptional expression in plants and animals. Although more than 1280 miRNAs in plant kingdom have been discovered thus far, only a handful of plant miRNAs has been intensively identified. The legume Medicago truncatula is a model system widely used to investigate gene transcription or post-transcriptional processes for arbuscular mycorrhiza development. In this study, we used a bioinformatics approach for ESTs (Expressed Sequence Tags)- and GSS (Genomic Survey Sequences)-wide prediction of novel miRNAs in M. truncatula. A total of 38 potential miRNAs were detected following a range of filtering criteria. After removal of 12 overlapping miRNAs that have already been deposited in miRNA Registry (Release 11.0), 26 miRNAs belonging to 15 families were found to be new. Using the newly identified miRNA sequences, we were able to further blast the M. truncatula mRNA database and detected 16 potential targets of miRNAs. Many miRNA targeted genes were predicted to encode transcription factors that regulate plant development, signaling, and metabolism. To verify the prediction of the miRNAs in M. truncatula, the qRT-PCR-based assay was performed. Our results showed that eight of the miRNA families displayed different patterns of expression in tissues. Additionally, expression of these miRNAs was up- or down-regulated by heavy metals (Hg, Cd, and Al). The above results implied that some of the miRNAs are involved in the regulation of development and plant response to heavy metal stress.