The effect of exercise on back muscle cross-sectional area (CSA), density, and strength was evaluated in patients sick-listed for subacute low back pain. Twenty-four patients were randomized into an exercise (n = 11) or a control (n = 13) group. Patients in the exercise group followed a biweekly exercise protocol for 15 weeks. Control patients received usual care. Muscle CSA and density were measured by computed tomography before and after intervention. Isokinetic test of back extensors was conducted simultaneously. Results showed a tendency to increased muscle CSA and density in patients in the exercise group, a significant decrease in muscle CSA at L4-L5 in control group patients, and a significant difference in change between groups in muscle CSA at L4-L5. Back extension strength increased in patients in the exercise group, but the improvement was not significant compared with control group patients. In conclusion, there was a tendency for reversal of muscle atrophy after exercise.