The objective of this study is to perform mini review on Hidden Markov Models (HMMs) which is recently important and popular among bioinformatics researchers and large no of software tools are based on this technique. The mathematical foundations of HMMs shall be considered first in brief manner and then the gene identification application. In the case of gene identification process, HMM basically resolve three basic problems: First is the evaluation problem, in this it computes the probability that a particular HMM will generates a given sequence of observations. Second is Decoding problem, in which it will uncover the most likely hidden state and Third is Learning problem, it is used to adjust the model parameter and train the HMM to find an optimal model. Evaluation problem can be solved by using Forward and Backward algorithm, Decoding problems are solved by using Viterbi algorithm and posterior decoding algorithm and then Learning problems are solved through Viterbi training algorithm and Baum-Welch algorithm. Finally, some limitations of the current approaches and future directions are also reported.