The evidence for effectiveness of different approaches is often diluted by the inclusion of heterogeneous groups and, in this case, lack of agreement over what constitutes 'acute' and 'chronic' back pain is a clear confounding factor. Although there are undoubtedly common issues in the approach to these problems, there are also clear differences. Closer definition of the problem and the development of specific and targeted outcome measures is required. In common with other areas of musculoskeletal rehabilitation, the evidence strongly supports exercise (except possibly in the rare case of true radicular back pain) and a cognitive behavioural approach to pain management. The variability of evidence in support of manipulation suggests that patient selection is important, but as yet those selection criteria are not clear. As is often the case, medical interventions are rarely submitted to evaluation in terms of functional outcome so, for procedures such as epidurals and facet joint injections, the jury is still out at the current time.