BackgroundAcupuncture is a popular form of complementary and alternative medicine (CAM), but it is not clear why patients do (or do not) follow acupuncturists’ treatment recommendations. This study aimed to investigate theoretically-derived predictors of adherence to acupuncture.MethodsIn a prospective study, adults receiving acupuncture for low back pain completed validated questionnaires at baseline, 2 weeks, 3 months, and 6 months. Patients and acupuncturists reported attendance. Logistic regression tested whether illness perceptions, treatment beliefs, and treatment appraisals measured at 2 weeks predicted attendance at all recommended acupuncture appointments.ResultsThree hundred twenty-four people participated (aged 18–89 years, M = 55.9, SD = 14.4; 70% female). 165 (51%) attended all recommended acupuncture appointments. Adherence was predicted by appraising acupuncture as credible, appraising the acupuncturist positively, appraising practicalities of treatment positively, and holding pro-acupuncture treatment beliefs. A multivariable logistic regression model including demographic, clinical, and psychological predictors, fit the data well (χ
2 (21) = 52.723, p < .001), explained 20% of the variance, and correctly classified 65.4% of participants as adherent/non-adherent.ConclusionsThe results partially support the dynamic extended common-sense model for CAM use. As hypothesised, attending all recommended acupuncture appointments was predicted by illness perceptions, treatment beliefs, and treatment appraisals. However, experiencing early changes in symptoms did not predict attendance. Acupuncturists could make small changes to consultations and service organisation to encourage attendance at recommended appointments and thus potentially improve patient outcomes.