Objectives: To prospectively investigate if the risk of chronic neck/shoulder pain is associated with work stress and job control, and to assess if physical exercise modifies these associations. Material and Methods: The study population comprised 29 496 vocationally active women and men in the Norwegian Nord-Trøndelag Health Study (HUNT Study) without chronic pain at baseline in 1984–1986. Chronic neck/shoulder pain was assessed during a follow-up in 1995–1997. A generalized linear model (Poisson regression) was used to calculate adjusted relative risks (RRs). Results: Work stress was dosedependently associated with the risk of neck/shoulder pain (p_trend < 0.001 in both sexes). The women and men who perceived their work as stressful “almost all the time” had multi-adjusted RRs = 1.27 (95% confidence interval (CI): 1.1–1.47) and 1.71 (95% CI: 1.46–2), respectively, referencing those with no stressful work. Work stress interacted with sex (p < 0.001). Poor job control was not associated with the risk of neck/shoulder pain among the women (RR = 1.04, 95% CI: 0.92–1.19) nor the men (RR = 1.09, 95% CI: 0.95–1.26). Combined analyses showed an inverse dose-dependent association between hours of physical exercise/week and the risk of neck/shoulder pain in the men with no stressful work (p_trend = 0.05) and among the men who perceived their work as “rarely stressful” (p_trend < 0.02). This effect was not statistically significant among the women or among men with more frequent exposure to work stress. Conclusions: Work stress is an independent predictor of chronic neck/shoulder pain and the effect is stronger in men than in women. Physical exercise does not substantially reduce the risk among the persons with frequent exposure to work stress.