To determine the effects of structured exercise programs on self-reported disability in older adults with knee osteoarthritis. A randomized, single-blind clinical trial lasting 18 months conducted at 2 academic medical centers. A total of 439 community-dwelling adults, aged 60 years or older, with radiographically evident knee osteoarthritis, pain, and self-reported physical disability. An aerobic exercise program, a resistance exercise program, and a health education program. The primary outcome was self-reported disability score (range, 1-5). The secondary outcomes were knee pain score (range, 1-6), performance measures of physical function, x-ray score, aerobic capacity, and knee muscle strength. A total of 365 (83%) participants completed the trial. Overall compliance with the exercise prescription was 68% in the aerobic training group and 70% in the resistance training group. Postrandomization, participants in the aerobic exercise group had a 10% lower adjusted mean (+/- SE) score on the physical disability questionnaire (1.71 +/- 0.03 vs 1.90 +/- 0.04 units; P<.001), a 12% lower score on the knee pain questionnaire (2.1 +/- 0.05 vs 2.4 +/- 0.05 units; P=.001), and performed better (mean [+/- SE]) on the 6-minute walk test (1507 +/- 16 vs 1349 +/- 16 ft; P<.001), mean (+/-SE) time to climb and descend stairs (12.7 +/- 0.4 vs 13.9 +/- 0.4 seconds; P=.05), time to lift and carry 10 pounds (9.1 +/- 0.2 vs 10.0 +/- 0.1 seconds; P<.001), and mean (+/-SE) time to get in and out of a car (8.7 +/- 0.3 vs 10.6 +/- 0.3 seconds; P<.001) than the health education group. The resistance exercise group had an 8% lower score on the physical disability questionnaire (1.74 +/- 0.04 vs 1.90 +/- 0.03 units; P=.003), 8% lower pain score (2.2 +/- 0.06 vs 2.4 +/- 0.05 units; P=.02), greater distance on the 6-minute walk (1406 +/- 17 vs 1349 +/- 16 ft; P=.02), faster times on the lifting and carrying task (9.3 +/- 0.1 vs 10.0 +/- 0.16 seconds; P=.001), and the car task (9.0 +/- 0.3 vs 10.6 +/- 0.3 seconds; P=.003) than the health education group. There were no differences in x-ray scores between either exercise group and the health education group. Older disabled persons with osteoarthritis of the knee had modest improvements in measures of disability, physical performance, and pain from participating in either an aerobic or a resistance exercise program. These data suggest that exercise should be prescribed as part of the treatment for knee osteoarthritis.