© 2017 IEEE.Predictive algorithms have become very important for handover (HO) management in mobile communications. In this regard, numerous techniques are being applied in order to obtain more accurate and robust methods. Markov chains (MC) are one of the most commonly used predictors since their easy implementation. In this paper, a threshold-based approach is introduced to common MC predictors in order to make predictions more accurate, since the probability is the main actor in a prediction process. The threshold value aims to prevent the predictor from making inaccurate predictions in case the probabilities of two or more states are very close. Results show that the proposed threshold-based method can improve the performance in terms of both prediction accuracy and signaling cost, specially for high randomness degrees, while also decreasing the number of inaccurate predictions.