There is uncertainty regarding how long the effects of acupuncture treatment persist after a course of treatment. We aimed to determine the trajectory of pain scores over time following acupuncture, using a large individual patient dataset from high quality randomized trials of acupuncture for chronic pain. The available individual patient dataset included 29 trials and 17,922 patients. The chronic pain conditions included musculoskeletal pain (low back, neck and shoulder), osteoarthritis of the knee and headache/migraine. We used meta-analytic techniques to determine the trajectory of post-treatment pain scores. Data on longer-term follow-up were available for 20 trials, including 6376 patients. In trials comparing acupuncture to no acupuncture control (wait-list, usual care, etc), effect sizes diminished by a non-significant 0.011 SD per 3 months (95% CI: -0.014 to 0.037, p = 0.4) after treatment ended. The central estimate suggests that about 90% of the benefit of acupuncture relative to controls would be sustained at 12 months. For trials comparing acupuncture to sham, we observed a reduction in effect size of 0.025 SD per 3 months (95% CI: 0.000 to 0.050, p = 0.050), suggesting about a 50% diminution at 12 months. The effects of a course of acupuncture treatment for patients with chronic pain do not appear to decrease importantly over 12 months. Patients can generally be reassured that treatment effects persist. Studies of the cost-effectiveness of acupuncture should take our findings into account when considering the time horizon of acupuncture effects. Further research should measure longer term outcomes of acupuncture.