Background Exercise is recommended for the treatment and management of low back pain (LBP) and the prevention of chronicity. Exercise adherence has proved to be only modest in intervention studies among people with musculoskeletal pain. Fear-avoidance beliefs (FABs) are known to affect exercise adherence.The purpose was twofold: to examine which bio-psycho-social factors contributed to exercise adherence during a 6-month neuromuscular exercise intervention among female healthcare workers with recurrent LBP, and to investigate how exercising affects FABs at 6 and 12 months’ follow-up.Methods In original RCT study, 219 healthcare workers aged 30–55 years with mild-to-moderate re-current non-specific LBP were allocated into: 1) exercise, 2) counselling, 3) combined exercise and counselling, and 4) control groups. In the present secondary analysis, groups 1 and 3 (exercise only and exercise+counselling) were merged to be exercisers and groups 2 and 4 were merged to be non-exercisers. Baseline variables of the exercise compliers (≥24 times over 24 weeks; n=58) were compared to those of the non-compliers (<1 time/week, 0–23 times; n=52). The effects of the exercise programme on FABs were analysed by a generalised linear mixed model according to the intention-to-treat principle (exercisers; n= 110 vs non-exercisers; n=109) at three measurement points (baseline, 6, and 12 months). A per-protocol analysis compared the more exercised to the less exercised and non-exercisers.Results A low education level (p=0.026), shift work (p=0.023), low aerobic (p=0.048) and musculoskeletal (p=0.043) fitness level, and high FABs related to physical activity (p=0.019) at the baseline contributed to lower exercise adherence. The exercise programme reduced levels of both physical activity- and work-related FABs, and there was a dose response: FABs reduced more in persons with better exercise adherence.Conclusion Healthcare workers who had lower education and fitness levels, worked shifts, and had high physical activity-related FABs had a lower adherence to the 6-month neuromuscular exercise programme. Exercising with good adherence reduced levels of FABs, which are known to be linked with prolonged LBP. In exercise interventions, motivational strategies should be targeted at persons with low education and fitness levels and high FABs in order to achieve better exercise adherence.