Research on complex systems has identified various aggregate relationships in phenomena that describe these systems. Travel length has been characterized by negative power distributions. Controversy, however, exists over whether mobility patterns can be modeled in terms of a simple power law (Lévy flight model) or that more complicated power laws (exponential power law, truncated Pareto) are required. This study concentrates on two issues: testing the validity of exponential power laws and truncated Pareto distributions in urban systems to describe aggregate mobility patterns, and examining differences in mobility patterns for different travel purposes. The article describes the results of an analysis of Global Positioning System (GPS) taxi trajectory data, collected in Guangzhou, China, to identify mobility patterns in the city. The least squares statistic is used to estimate the parameters of the distribution functions. Results suggest that a fusion of functions, based on an exponential power law and a truncated Pareto distribution, represents the travel time distribution best. Moreover, the findings of this study indicate different mobility patterns to exist for different travel purposes.