42 patients with undisplaced, proximal humeral fractures were randomly assigned into two groups in order to compare the results of instruction to the patient for independent exercises and conventional physiotherapy. No differences appeared between the groups subjectively or as regards functional parameters at 1 as well as at 3 months after the fracture and at a follow up more than 1 year later. As no disadvantage could be found instructions to the patients should in these cases be regarded as a satisfactory after-treatment.