An operant behavioural and time-contingent graded exercise therapy program was developed to improve functional ability irrespective of pain experience in patients with chronic shoulder complaints. The clinical effectiveness of graded exercise therapy compared to usual care was evaluated in a randomised clinical trial. Assessments were carried out before and after 12 weeks of treatment. Performance of daily activities was measured by two outcome measures: the main complaints instrument and the Shoulder Disability Questionnaire (SDQ). Patients were eligible for participation if they had suffered from shoulder complaints for at least three months. Patients suffering from systemic diseases, referred pain or severe biomedical or psychiatric disorders were excluded. Patients (n = 176) were randomised and allocated either to graded exercise therapy (n = 87) or usual care (n = 89). Graded exercise therapy led to greater improvement in the performance of daily activities than usual care. However, only mean differences between groups in performance of activities related to the main complaints reached statistical significance (p = 0.049; 95% CI 0.0 to 15.0). The observed beneficial effects were considered to be small to moderate (calculated effect sizes: 0.30 for the main complaints instrument and 0.07 for the SDQ). Subgroup analysis showed larger improvements on the mean complaints instrument in patients not reporting pain reduction over time. Graded exercise therapy seems to be less effective in restoring performance of daily activities as assessed by the SDQ in patients showing a painful arc during physical examination. Results showed that graded exercise therapy is more effective in restoring the ability to daily activities in patients with chronic shoulder complaints than usual care, although beneficial effects are small.