The nonspecific effects of acupuncture are well documented; we wished to quantify these factors in osteoarthritic (OA) pain, examining needling, the consultation, and the practitioner. In a prospective randomised, single-blind, placebo-controlled, multifactorial, mixed-methods trial, 221 patients with OA awaiting joint replacement surgery were recruited. Interventions were acupuncture, Streitberger placebo acupuncture, and mock electrical stimulation, each with empathic or nonempathic consultations. Interventions involved eight 30-minute treatments over 4 weeks. The primary outcome was pain (VAS) at 1 week posttreatment. Face-to-face qualitative interviews were conducted (purposive sample, 27 participants). Improvements occurred from baseline for all interventions with no significant differences between real and placebo acupuncture (mean difference -2.7 mm, 95% confidence intervals -9.0 to 3.6; P=.40) or mock stimulation (-3.9, -10.4 to 2.7; P=.25). Empathic consultations did not affect pain (3.0mm, -2.2 to 8.2; P=.26) but practitioner 3 achieved greater analgesia than practitioner 2 (10.9, 3.9 to 18.0; P=.002). Qualitative analysis indicated that patients' beliefs about treatment veracity and confidence in outcomes were reciprocally linked. The supportive nature of the trial attenuated differences between the different consultation styles. Improvements occurred from baseline, but acupuncture has no specific efficacy over either placebo. The individual practitioner and the patient's belief had a significant effect on outcome. The 2 placebos were equally as effective and credible as acupuncture. Needle and nonneedle placebos are equivalent. An unknown characteristic of the treating practitioner predicts outcome, as does the patient's belief (independently). Beliefs about treatment veracity shape how patients self-report outcome, complicating and confounding study interpretation.