Much of the research into return to work following rehabilitation for low back pain in the literature reflects work done in those employed. Unemployment is a consequence of chronic low back pain which has considerable health and economic consequences for the individual and society. This paper describes an occupationally orientated rehabilitation programme for long-term unemployed people (mean duration of unemployment 38.9 months). The aim of the project was to identify factors which predict return to work and progress towards employment. Eight six subjects underwent a pain management rehabilitation programme incorporating vocational focussing and advice, subjects were followed up at 6 months to determine work status. At follow-up 38.4% of subjects were employed and another 23% were in voluntary work, or education/training. There were no significant differences on presenting characteristics between those who returned to work and those who did not. Subjects were divided into those who made positive progress (work, education/training or voluntary work) and those who did not (remained unemployed, dropped out of the programme or lost to follow up). Those who failed to make positive progress were characterised by longer duration of unemployment and higher scores on somatic anxiety and depression. A predictive model was able to identify 80% of those who failed to make progress but prediction of those achieving a positive outcome was poor (44% correct prediction). The factors predicting return to work in unemployed people with low back pain differs from the employed, the need for employment skills training and a vocational focus to rehabilitation are highlighted.