The invention relates to the field of network user behavior analysis and particularly relates to a communication network user calling behavior prediction method and device based on maximum likelihood.The communication network user calling behavior prediction method based on maximum likelihood comprises the steps of carrying out statistics on CDR data of a calling number at different time periods,wherein the CDR data comprises call frequencies and call duration, and recording the maximum call frequency and maximum call duration of each time period; establishing a call frequency-probability density distribution curve and a call duration-probability density distribution curve of the time period according to the call frequency and the call duration of the calling number in each time period;and predicting the call frequency and call duration of the calling number in each time period according to the call frequency-probability density distribution curve and the call duration-probability density distribution curve. The communication network user call behavior prediction device based on maximum likelihood comprises a statistics module, an establishment module and a prediction module. According to the method and the device, user calling behaviors can be predicted.