AbstractAs the ability to collect data continues to outstrip the ability to process and analyze it, the age‐old paradigm of store‐and‐process is becoming untenable. Finding one or two interesting items in the midst of many possible signals depends on context which often changes over time. A new way to interact with data is needed to handle some of these challenges. The streaming data model is one such approach. In this article, we present the streaming data model as well as two approaches to designing algorithms to handle streaming data. The first, the single processor method, traces its heritage to database models. The second, the multi‐processor method, is more aligned with signal processing algorithms. We will close with a critique of the current approaches and some of the statistical challenges that streaming data pose. WIREs Comp Stat 2011 3 22–29 DOI: 10.1002/wics.130This article is categorized under: Algorithms and Computational Methods > Algorithms Data: Types and Structure > Streaming Data Statistical Learning and Exploratory Methods of the Data Sciences > Streaming Data Mining