In order to identify the high conditional complexity in source code, a novel approach based on statistical analysis is proposed. According to the statistical analysis of two software metrics which are Method McCabe's Cyclomatic Complexity (MMCC) and Method Average McCabe's Cyclomatic Complexity Per Code Line (MAMCC) in a large number of projects, the probability density functions and cumulative distribution functions for describing distributions of these two metrics are obtained. Moreover, a model for identifying high conditional complexity is built by choosing reasonable threshold of these me...