A three-year study examined changes in N95 filtering-facepiece respirator (FFR) fit at six-month intervals and the relationship between fit and changes in weight for 229 subjects. During each visit, subjects performed a total of nine fit tests using three samples of the same FFR model. Inward leakage and filter penetration were measured for each donned respirator to determine face seal leakage (FSL). A total of 195 subjects completed the second visit and 134 subjects completed all seven visits. Acceptable fit was defined as 90th percentile FSL ≤ 5% and at least one fit factor ≥ 100. An unacceptable fit was observed for 14, 10, 7, 12, 15, and 16% of subjects on Visits 2–7, respectively. The predicted risk of an unacceptable fit increased with increasing length of time between fit tests, from 10% at Year 1 to 20% at Year 2 and to 25% at Year 3. Twenty-four percent of subjects who lost ≥ 20 lb had an unacceptable fit; these percentages ranged from 7–17% for subjects with lower weight losses or any degree of weight gain. Results support the current OSHA requirement for annual fit testing and suggest that respirator users who lose more than 20 lb should be re-tested for respirator fit.