Musculoskeletal symptoms are frequent in primary care, and back pain and osteoarthritis are the commonest causes of disability in the developed world. Effective primary care could make a real difference to population levels of pain and disability. Traditional approaches, which see musculoskeletal pain only as a guide to underlying pathology, underestimate the need to assess and manage pain and disability in their own right. This paper presents the rationale for new models of diagnosis and care. These include: placing symptoms and their impact on daily life at the centre of primary care management of common musculoskeletal syndromes; a focus on the person with co-morbidity rather than on neatly parcelled distinctive single musculoskeletal diagnoses; promotion of positive messages such as the safety of activity and the efficacy of simple treatments; inclusion of the psychological and social context in management of musculoskeletal conditions; supporting patients to live and participate in daily life despite their pain or disability; supporting and training health care professionals other than doctors to provide a ‘gatekeeping’ role for these conditions; evaluation of changes in the organisation of care for patients with musculoskeletal problems. Evidence for the effectiveness of such new models is uneven. The paper reviews other research directions, such as qualitative research into patient perceptions and priorities, and the idea that non-specific treatment effects (eg, acupuncture) are often bigger than specific effects and could be harnessed to improve clinical care. Natural history studies remain important for understanding causation: prevalence is already near lifetime maximum by young adulthood for example and we do not know why. Musculoskeletal medicine is the Cinderella of primary care and primary care research. Knowledge, training and enthusiasm lags behind other priority areas for primary care clinicians. Yet the potential benefits, which primary care could achieve, are enormous.