This study was performed to estimate the discriminatory power of multiple combinations of risk indicators for the occurrence and recurrence of low-back trouble (LBT) in workers. Two categories of LBT provided groups for discrimination; 1) the presence or absence of LBT history, and 2) three patterns of recurrence characterized by the number of episodes (isolated, periodic, chronic). The risk indicators comprised data reflecting occupational and leisure demands on the back, measures of lumbar sagittal mobility, and anamnestic features of the first episode. Discriminant analysis was the statistical procedure used. The results showed that it was possible to find linear combinations of the discrimination variables that successfully allocated around two-thirds of the sample to the correct group. The presence of a history of LBT was predicted by the combined effect of increasing age and adult sports participation, but only in females did a heavier job contribute to such prediction. A reduction in risk was associated with lumbar flexibility and sports participation at school. Chronic LBT was more accurately identified than the two other groups; increasing age, a long initial spell, and an onset early in life were associated with increased likelihood of chronicity, while a report of symptoms being relieved by sitting reduced this risk. It is concluded that the occurrence and recurrence of LBT are related to combinations of risk indicators, and that it is imperative to consider the interactive effect of a multiplicity of factors in epidemiologic studies.