Graduation date: 2007Generalizability theory was used to examine sources of variability in daily physical activity levels of secondary students with and without developmental disabilities (DD), and to determine minimum number of days required for monitoring their typical physical activity levels. Sixteen participants with DD (M = 16.7 years, SD = 2.7 years), and thirty-one (M = 12.3 years, SD = 0.5 years) children without DD participated in the study. They wore two pedometers and two accelerometers during 5 weekdays and 4 weekend days. Sources of variability were examined using a two-facet fully crossed design. Twelve separate two-way ANOVAs were employed for each population, physical activity device, and measurement periods (weekday, weekends, and weekdays and weekends combined). For participants with DD, variance components of the person, and the person by day interaction were the primary sources of variability in daily physical activity for pedometers and accelerometers across weekdays, weekend days, and weekdays and weekends combined. To determine the typical physical activity level with generalizability coefficients of .80, at least 4, 6, and 8 days of measurement using a pedometer were required during weekdays, weekend days, and weekdays and weekends combined, respectively. Using an accelerometer, at least 4 days of measurements were needed across weekdays, weekends, and weekdays and weekends combined. For participants without DD, the primary sources of variability during weekdays and weekends were related to variance components of the persons and the person by day interaction for both pedometers and accelerometers. When weekdays and weekends were combined, relatively large percentages of variability were associated with the residual, indicating three way interaction, plus unexplained error. Using one pedometer, to achieve generalizability coefficients of .80 in the measurement of daily physical activity, a minimum number of 5 and 9 days of measurements during weekday and weekends were estimated, respectively. Using one accelerometer, at least 4 days and 14 days of monitoring physical activity were required during weekdays, and weekdays and weekends combined, respectively. However, an estimation of typical physical activity levels during weekdays and weekends combined, using one pedometer as well as during weekends using one accelerometer was unfeasible due to the number of days required for measurement.