Purpose To determine the time-wise effect of specific resistance training on neck pain among industrial technicians with frequent neck pain symptoms. Methods Secondary analysis of a parallel-group cluster randomized controlled trial of 20 weeks performed at two large industrial production units in Copenhagen, Denmark. Women with neck pain >30 mm VAS (N = 131) were included in the present analysis. The training group (N = 77) performed specific resistance training for the neck/shoulder muscles three times a week, and the control group (N = 54) received advice to stay active. Participants of both groups registered neck pain intensity (0–100 mm VAS) once a week. Results Neck pain intensity was 55 mm (SD 23) at baseline. There was a significant group by time interaction for neck pain (F-value 2.61, P<0.001, DF = 19). Between-group differences in neck pain reached significance after 4 weeks (11 mm, 95% CI 2 to 20). The time-wise change in pain showed three phases; a rapid decrease in the training group compared with the control group during the initial 7 weeks, a slower decrease in pain during the following weeks (week 8–15), and a plateau during the last weeks (week 16–20). Adherence to training followed a two-phase pattern, i.e. weekly participation rate was between 70–86% during the initial 7 weeks, dropping towards 55–63% during the latter half of the training period. Conclusion Four weeks of specific resistance training reduced neck pain significantly, but 15 weeks is required to achieve maximal pain reduction. The time-wise change in pain followed a three-phase pattern with a rapid effect during the initial 7 weeks followed by a slower but still positive effect, and finally a plateau from week 15 and onwards. Decreased participation rate may explain the decreased efficacy during the latter phase of the intervention.