Mobile phones are widely used by people in modern society, the records of which provide us with an essential proxy for the understanding of human behaviors. Previous empirical results have already shown that some temporal, even spatial, characteristics of many human activities follow power law instead of Poisson statistics. In this paper, we investigate the statistics of human mobile phone communication patterns and their mobility patterns based on mobile phone records. The results show that the distributions of related person and the distributions of inter-event time for four different behaviors all follow power law distribution with different values of exponent. In addition, we have a comparative analysis of human behavior in four different periods of a day, and find out that most of individuals are more active in the afternoon and less active at night in aspects of the diversities of persons they reached and the range of their activities.