Accelerometry data were collected on approximately 7000 participants ages 6y and over in the National Health and Nutrition Examination Survey 2003-2004. This presentation describes data cleaning and analytic decisions applied to these data. Topics addressed include detection of monitor malfunction, definition of valid days and persons for analysis, selection of intensity and bout length criteria for moderate and vigorous activity, and requirements for handling a 3 gigabyte data file. Implications of various decisions are explored and demonstrated.