Background: Musculoskeletal and chronic pain complaints account for large numbers of health care services and a disproportionate amount of expense. However, there is substantial doubt about the effectiveness of many tests and treatments in those clinical areas. Objectives: To critically appraise the literature on testing and treatment of musculoskeletal and chronic pain problems. To provide evidence-based guidance for testing and treatment. To suggest improvements in research design and execution in these areas. Methods: Researchers posed answerable clinical questions, searched MEDLINE, Cochrane databases, EMBASE, PEDro, CINAHL and other sources, screened thousands of studies and reviews, and critically appraised selected studies using methodologies adapted from the Cochrane back group, other Cochrane approaches and USPSTF. Results: This structured appraisal of individual studies revealed considerable risk of bias. Reviewers found widespread design and execution problems such as use of subjective outcomes; small group size; lack of case/diagnostic definitions; non-standard or black box interventions; no true control groups; crossover among groups; unacceptable drop out rates; short-term follow-up only; and failure to assess harms. Known variables affecting these complaints, such as SES, education, job control and latitude, job and task satisfaction, satisfaction with care, locus of control and litigation, were seldom included as variables. Function was seldom used as an outcome. Most systematic reviews suffered from incomplete searches, nonstandard study grading, selective use of studies, and inadequate or incorrect interpretation of results. Results were often heterogeneous. Conclusions: These errors reduce the likelihood that the resulting conclusions or guideline recommendations would be consistent or reproducible. This presentation will suggest an agenda for research to produce more robust evidence to support clinical practice and patient information. Functional outcomes and relative efficiency are primary concerns.