Objectives To determine the impact of a 12-week ergonomic/exercise programme compared with an ergonomic/health education programme on the development of neck pain in office workers over 12 months. Methods This cluster-randomised trial prospectively recruited office workers from public and private organisations. Only non-neck pain cases at baseline were included (n=484). All participants received an ergonomic workstation review then randomly allocated to receive a neck/shoulder progressive exercise programme (20 min, 3 x/week; intervention group) or health education sessions (60 min, 1 x/week; active control) for 12 weeks. Generalised estimating equations evaluated group differences in the point prevalence of neck pain cases (defined as those with a neck pain score of >= 3 on a 0 (no pain) to 9 (worst pain) scale) over time (3, 6, 9 and 12 months) with cumulative incidence of neck pain cases evaluated descriptively. Results While no significant group x time interaction was evident, the 12-month point prevalence of neck pain cases in the intervention group (10%) was half that of the active control group (20%) (adjusted OR 0.46, 95% CI 0.21 to 1.01, p=0.05). Lower cumulative incidence of neck pain cases was observed in the intervention (17%) compared with active control group (30%) over the 12 months. Conclusions A combined ergonomics and exercise intervention may have more benefits in preventing neck pain cases in office workers than an ergonomic and health education intervention. Group differences were modest and should be interpreted with caution when considering strategies for primary prevention of neck pain in the office worker population.