We describe a method to derive safe approximations for data dependences in programs with pointers and structures. In our approach, alias information and reaching definitions' information at each program point is simultaneously covered by a single representation, called A/D graphs. We perform a single-pass data dependence analysis for a class of restricted imperative languages by solving a monotone data flow system, based on A/D graphs. Advantages of our method are improved accuracy, economical storage use, and reduced analysis time.