We investigated muscle strength, aerobic power, and occupational and leisure-time physical loading as predictors of back pain in a 5-year follow-up study. A cohort of 456 adults aged 25, 35, 45 and 55 years, free of back pain, participated in measurements of anthropometric characteristics, aerobic power and muscle strength characteristics at baseline. The subjects' levels and types of physical activity and occupational physical loading were also determined. At 5 years after the baseline examinations 356 of these subjects (78.1%) were reached by mail, and 262 of them (73.6%) properly completed and returned a questionnaire including a detailed back pain history for the 5 years following the baseline measurements. Of this number 56 subjects (21%) who reported back pain (> 30 on a scale from 0 to 100) and functional impairment during the 5-year follow-up composed the marked back pain group. Other subjects (n = 71, 27%) noting lesser symptoms were included in the mild back pain group; 135 subjects (52%) reported having had no back pain. The subjects with marked back pain were on average taller than the subjects without back pain, while no such difference was found in body mass. Heavy occupational musculoskeletal loading (P = 0.005) and high general occupational physical demands (P = 0.036) predicted future back pain. Leisure-time physical activity, aerobic power or muscle strength characteristics were not predictive of future back pain.