Aims: The aims of this study were to test whether a multi-faceted intervention effective for low back pain was effective for physical capacity, work demands, maladaptive pain behaviours, work ability and sickness absence due to low back pain. Methods: A stepped wedge cluster randomised, controlled trial with 594 nurses' aides was conducted. The intervention lasted 12 weeks and consisted of physical training (12 sessions), cognitive behavioural training (two sessions) and participatory ergonomics (five sessions). Occupational lifting, fear avoidance, physical exertion, muscle strength, support from management, work ability and sickness absence due to low back pain were measured every 3 months. Before and after the intervention we measured physical capacity, kinesiophobia and need for recovery. Linear mixed models adjusted for baseline values of the outcome were used to estimate the effect. Results: Significant reduction in occupational lifting (-0.35 (95% confidence interval -0.61 to -0.08)), and improvement in two measures of fear avoidance ((-0.75 (95% confidence interval -1.05 to -0.45) and -0.45 (95% confidence interval -0.80 to -0.11)) were found for the intervention group compared to the control. There were no significant effects on physical exertion, muscle strength, support from management, work ability or sickness absence due to low back pain. After the intervention, significant increased physical capacity and improvements in kinesiophobia were found, but no change in need for recovery. Conclusions: The intervention was significantly effective for physical work demands and maladaptive pain behaviours, but not for work ability and sickness absence due to low back pain. To improve work ability or reduce sickness absence due to low back pain more specific interventions should probably be developed.