An elevator reliability prediction method based on an unbiased grey fuzzy Markov chain model is presented. When the elevator breaks down, the fault code is extracted, recorded and the historical datais obtained. The unbiased GM (1, 1) model is established according to the historical data, and the residual error of the model is checked. The residual value interval is divided into m states, fuzzy function is constructed by fuzzy classification theory, fuzzy state probability vector of each data point is calculated, and the state of each data point is determined according to the maximum membership principle. The Markov state transition matrix is established to predict the fuzzy vector of the fault next time, and the membership state of the predicted data is determined according to the maximum membership principle. The invention predicts the reliability of the elevator, obtains the time interval trend of the occurrence of a certain fault through an unbiased GM (1, 1), approaches the datafluctuation trend by utilizing the anti-interference property of the fuzzy Markov chain, and slides the Markov chain, so that the reliability prediction result of the elevator is more accurate.