Shoulder pain is common in primary care and has an unfavourable outcome in many patients. Information about predictors of outcome is scarce and inconsistent. The objective of this study was to develop clinical prediction rules for calculating the absolute risk of persistent shoulder symptoms for individual patients, 6 weeks and 6 month after the first consultation in general practice. A prospective cohort study with 6 months follow-up was carried out in three geographic areas in The Netherlands. In this study, 587 patients with a new episode of shoulder pain were included. The main outcome measure was persistent symptoms at 6 weeks and 6 months, perceived by the patient. Potential predictors included the results of a physical examination, sociodemographic variables, disease characteristics (duration of symptoms, pain intensity, disability and comorbidity), physical activity, physical workload and psychosocial factors. Response rates to the follow-up questionnaires were 83% at 6 weeks and 92% at 6 months. A longer duration of symptoms, gradual onset of pain and high pain severity at presentation were consistently associated with persistent symptoms at 6 weeks and 6 months. The discriminative validity of our prediction rules was satisfactory with area under the curves of 0.74 (95% CI 0.70, 0.79) at 6 weeks and 0.67 (95% CI 0.63, 0.71) at 6 months. The performance of our rules needs to be tested in other populations of patients with shoulder pain to enable valid and reliable use of the rules in everyday clinical practice.