This study examined the number of days of global positioning system (GPS) monitoring needed to measure attributes of an individual's routine activity space. Multiple alternative activity space representations (cumulative, mean daily), measures (kernel density, route buffer, convex hull), and attributes (area size, supermarkets, fast food restaurants, parks) were examined. Results suggested wide variability in required GPS days to obtain valid estimates of activity space attributes (1–23 days). In general, fewer days were needed for mean daily activity space representations, kernel density measures, and densities of environmental exposures (vs. counts). While kernel density measures reliably estimated between-person differences in attributes after just a few days, most variability in environmental attributes for convex hull and route buffer measures was within-person. Based on these results, a minimum of 14 days of valid GPS data is recommended to measure activity spaces. •In contrast to current norm of 7 days, at least 14 days of valid GPS monitoring are needed.•Days for valid estimates varied by activity space representation, measure, and environmental attribute.•Kernel density measures achieved reliable estimates after a few days.•Reliable estimates were not found for convex hull and route buffer measures.•Activity space attributes, when not weighted by time, vary considerably day-to-day.