PURPOSE: To review the efficacy of nonmedicinal, noninvasive therapies in hip and knee osteoarthritis. DATA SOURCES: Search of English-language literature from 1966 through 1993 using MEDLINE by cross-referencing "osteoarthritis" (therapy subheadings) with "controlled trial," "comparative study," or "trial(s)." STUDY SELECTION: Fifteen controlled trials of diathermy (deep heat), exercise, acupuncture, transcutaneous electrical nerve stimulation, topically applied capsaicin, low-energy laser, and pulsed electromagnetic fields were found. No experimental studies of superficial heat and cold, orthotic devices, vibration, or weight loss were identified. RESULTS: Exercise reduces pain and improves function in patients with osteoarthritis of the knee. No support exists in the literature for pre-exercise ultrasound treatment. Single, well-designed studies suggest that topically applied capsaicin and laser treatment reduce pain associated with knee osteoarthritis. Data on the other three therapies were sparse (transcutaneous electrical nerve stimulation, pulsed electromagnetic fields) or inconsistent (acupuncture). CONCLUSIONS: More data are needed to determine the optimal exercise regimen for treating knee osteoarthritis and to evaluate the role of topical capsaicin, laser therapy, acupuncture, transcutaneous electrical nerve stimulation, and pulsed electromagnetic fields. No data specifically address the role of any of these therapies in hip osteoarthritis.