This study examined long-term effects of a tailored behavioural treatment protocol (TBT), as compared with an exercise based physical therapy protocol (EBT). One-hundred and twenty-two patients who, due to persistent musculoskeletal pain, consulted physical therapists in primary care were originally randomized to either of the two conditions. Follow-up assessments two-year post-treatment were completed by 65 participants. According to per-protocol analyses, short-term effects were maintained in both groups for the primary outcome, pain-related disability. The TBT-group reported lower disability levels compared with the EBT-group. Intention-to-treat analyses (ITT) conveyed similar results. Secondary outcomes of pain intensity, pain control, and functional self-efficacy were maintained over the 2-year post-treatment, but previous group differences were levelled out according to the most conservative method of ITT. Fear of movement/(re)injury increased in the EBT-group, and EBT participants reported higher fear of movement/(re)injury two years post-treatment compared to TBT. The study supports tailoring of treatments in concordance with patients' needs and preferences of activity goals and functional behavioural analyses including predictors of pain-related disability, for successful immediate outcomes and their maintenance in the long run. Exercise-based treatments resulted in somewhat smaller immediate treatment effects but had similar maintenance of effects over the 2-year follow-up period.