Computational prediction of microRNAs is one of the most important approaches in microRNA studies. However, validation of the predicted microRNAs’ precise sequences is essential for further studies on microRNA biogenesis, evolution and functions. Here, we report a highly efficient method of determining the precise sequences of computationally predicted miRNAs, which method combines a miRNA-enriched library preparation, two specific 3’ and 5’ miRNA RACE (miR-RACE) reactions, and sequence-directed cloning. miR-RACE has the potential to overcome the major disadvantage of computational miRNA prediction methods in that they can not predict the precise sequence of miRNAs, which could make the bioinformatic prediction of miRNAs more powerful and accurate. The efficiency of this method can be reflected from the precise sequence validation of the miRNAs computationally predicted in citrus, apple, and some other fruit crops. Our on-going research indicates that miR-RACE can also be very effective in the verification of sequences of some doubtful miRNAs obtained by deep sequencing of small RNA libraries. The protocol of miR-RACE is a rapid one to be executed and can be completed in 2-3 days.