In this paper, we investigate a new approach to process queries in data stream applications. We show that reference locality characteristics of data streams could be exploited in the design of superior and flexible data stream query processing techniques. We identify two different causes of reference locality: popularity over long time scales and temporal correlations over shorter time scales. An elegant mathematical model is shown to precisely quantify the degree of those sources of locality. Furthermore, we analyze the impact of locality-awareness on achievable performance gains over traditional algorithms on applications such asMAX-subset approximate sliding window join and approximate count estimation. In a comprehensive experimental study, we compare several existing algorithms against our locality-aware algorithms over a number of real datasets. The results validate the usefulness and efficiency of our approach.