OBJECTIVES: The objective of this study is to determine if exercise increases joint symptoms in older adults with a history of arthritis or produces symptoms in older adults without such history. In addition, we examine whether joint symptoms explain the large observed variation in strength gain in older adults undergoing vigorous strength training exercise, and report the incidence of musculoskeletal injuries upon initiation of an exercise program.DESIGN: A population‐based, single blinded, randomized controlled trial with three exercise groups and one control group.SETTING: A large urban health maintenance organization.PARTICIPANTS: Older men and women (N = 105) aged 68 to 85, with leg strength below the 50th percentile for their age, sex, height, and weight and without neuromuscular disease or active cardiovascular disease.INTERVENTIONS: Supervised exercise in 1‐hour sessions, three times each week, for 24 to 26 weeks. One exercise group did strength training (ST) using weight machines (n = 25); another group did endurance training (ET) using stationary cycles (n = 25); and the third group did combined strength training and endurance training (ST+ET) (n = 25). The control group (n = 30) received no intervention.MEASUREMENTS: Strength was measured at the ankle, knee, hip, and elbow using an isokinetic dynamometer. Joint symptoms were rated on a 6‐point scale (0 = none, 5 = severe). Arthritis severity was based on self‐reported use of arthritis medication. Health status was measured with sub‐scales of the SF‐36 and Sickness Impact Profile (SIP).RESULTS: Joint symptoms fluctuated over time in all exercise groups, but they did not improve or worsen significantly in any group. The physical dimension of the SIP and SF‐36 subscale scores, including Bodily Pain Scores, did not change over time in any group. Subjects with arthritis and joint symptoms gained as much strength with strength training as did subjects without joint symptoms. Adjustment for age, gender, baseline strength, adherence, and exercise group did not affect this finding. The rate of minor musculoskeletal injuries was 2.2 injuries per 1000 exercise hours.CONCLUSIONS: Moderate intensity stationary cycle exercise and vigorous intensity strength training do not appear to produce or exacerbate joint symptoms in older adults. Joint symptoms did not explain the large variation in gains in strength in older adults participating in a standardized strength training exercise program. Musculoskeletal injuries occurred relatively infrequently, and no major injuries occurred. In evaluating joint pain that occurs in older adults in well regulated exercise programs, clinicians should consider other etiologies before attributing pain to exercise per se.J Am Geriatr Soc 4