At present, there is an increasing international trend towards evidence-based health care. The field of low back pain (LBP) research in primary care is an excellent example of evidence-based health care because there is a huge body of evidence from randomized trials. These trials have been summarized in a large number of systematic reviews. This paper summarizes the best available evidence from systematic reviews conducted within the framework of the Cochrane Back Review Group on non-invasive treatments for non-specific LBP. Data were gathered from the latest Cochrane Database of Systematic Reviews 2005, Issue 2. The Cochrane reviews were updated with additional trials, if available. Traditional NSAIDs, muscle relaxants, and advice to stay active are effective for short-term pain relief in acute LBP. Advice to stay active is also effective for long-term improvement of function in acute LBP. In chronic LBP, various interventions are effective for short-term pain relief, i.e. antidepressants, COX2 inhibitors, back schools, progressive relaxation, cognitive–respondent treatment, exercise therapy, and intensive multidisciplinary treatment. Several treatments are also effective for short-term improvement of function in chronic LBP, namely COX2 inhibitors, back schools, progressive relaxation, exercise therapy, and multidisciplinary treatment. There is no evidence that any of these interventions provides long-term effects on pain and function. Also, many trials showed methodological weaknesses, effects are compared to placebo, no treatment or waiting list controls, and effect sizes are small. Future trials should meet current quality standards and have adequate sample size.