Prognostic studies on neck pain are scarce and are typically restricted to short-term follow-up only. In this prospective cohort study, indicators of short- and long-term outcomes of neck pain were identified that can easily be measured in general practice. Patients between 18 and 70 years of age, suffering for at least 2 weeks from neck pain were recruited by 42 general practitioners (GPs). Perceived recovery, pain intensity and neck dysfunction after 7 and 52 weeks were considered as outcome measures. Indicators of prognosis were identified by means of logistic regression analyses (perceived recovery) and linear regression analyses (pain intensity and neck dysfunction). In total, 183 patients were included. After 1 year, 63% had recovered. The prognostic models showed differences between short- and long-term indicators. At the short term, besides the baseline values of the respective outcome measurements, only older age (> or =40) and concomitant low back pain and headache were associated with poor outcome. At the long term, in addition to age and concomitant low back pain, previous trauma, a long duration of neck pain, stable neck pain during the 2 weeks prior to baseline measurement, and previous neck pain predicted poor prognosis. The predictive power of the models was weak: the explained variance (R(2)) varied from 24 to 36%. Patient history and physical examination give GPs little handholds to predict the prognosis for patients with sub-acute and chronic neck pain. A few indicators of a less favourable prognosis of neck pain were identified, of which older age and concomitant low back pain was the most consistent.