An analysis of supply and maintenance data collected over seven months at Wright-Patterson Air Force Base (AFB). One squadron of B-52s and miscellaneous aircraft generated 10,965 maintenance actions (for 3,220 items), 51 percent being repairs on base, 4 percent bench-checked serviceable, and 45 percent not reparable this station (NRTS). Half of the items had only one demand in the 213 days. Net repair cycles averaged 10.7, 6.6, and 6.8 days for repaired, bench-check, and NRTS, respectively. Including maintenance and parts delays adds 3.5 days to repaired and NRTS actions and 1 day to bench-checks. Although demand levels for most individual items are insensitive to chance variations in repair cycle times, the aggregate demand level for the group is responsive to even small changes in the average group cycle time. NRTS actions account for a third of the on-base net processing time, yet this time is excluded from the stock-level computations. Including it would increase the aggregate demand level for the 3,220 items by 37 percent. An average repair cycle time can be used to estimate the demand level for a group of items with reasonable accuracy. By expressing this level as a ratio of stock per item, it is possible to determine the effect of changes in repair cycles or in average demand for different groups of items.