OBJECTIVES: Chronic pain is common and use of long-term opioid therapy for chronic pain has increased dramatically. This report reviews the current evidence on effectiveness and harms of opioid therapy for chronic pain, focusing on long-term (≥1 year) outcomes. DATA SOURCES: A prior systematic review (searches through October 2008), electronic databases (Ovid MEDLINE, Scopus, and the Cochrane Libraries January 2008 to August 2014), reference lists, and clinical trials registries. REVIEW METHODS: Using predefined criteria, we selected randomized trials and comparative observational studies of patients with cancer or noncancer chronic pain being considered for or prescribed long-term opioid therapy that addressed effectiveness or harms versus placebo, no opioid use, or nonopioid therapies; different opioid dosing methods; or risk mitigation strategies. We also included uncontrolled studies ≥1 year that reported rates of abuse, addiction, or misuse, and studies on the accuracy of risk prediction instruments for predicting subsequent opioid abuse or misuse. The quality of included studies was assessed, data were extracted, and results were summarized qualitatively. RESULTS: Of the 4,209 citations identified at the title and abstract level, a total of 39 studies were included. For a number of Key Questions, we identified no studies meeting inclusion criteria. Where studies were available, the strength of evidence was rated no higher than low, due to imprecision and methodological shortcomings, with the exception of buccal or intranasal fentanyl for pain relief outcomes within 2 hours after dosing (strength of evidence: moderate). No study evaluated effects of long-term opioid therapy versus no opioid therapy. In 10 uncontrolled studies, rates of opioid abuse were 0.6 percent to 8 percent and rates of dependence were 3.1 percent to 26 percent in primary care settings, but studies varied in methods used to define and ascertain outcomes. Rates of aberrant drug-related behaviors ranged from 5.7 percent to 37.1 percent. Compared with nonuse, long-term opioid therapy was associated with increased risk of abuse (one cohort study), overdose (one cohort study), fracture (two observational studies), myocardial infarction (two observational studies), and markers of sexual dysfunction (one cross-sectional study), with several studies showing a dose-dependent association. One randomized trial found no difference between a more liberal opioid dose escalation strategy and maintenance of current dose in pain or function, but differences between groups in daily opioid doses at the end of the trial were small. One cohort study found methadone associated with lower risk of mortality than long-acting morphine in a Veterans Affairs population in a propensity adjusted analysis (adjusted HR 0.56, 95 percent CI 0.51 to 0.62). Estimates of diagnostic accuracy for the Opioid Risk Tool were extremely inconsistent and other risk assessment instruments were evaluated in only one or two studies. No study evaluated the effectiveness of risk mitigation strategies on outcomes related to overdose, addiction, abuse, or misuse. Evidence was insufficient to evaluate benefits and harms of long-term opioid therapy in high-risk patients or in other subgroups. CONCLUSIONS: Evidence on long-term opioid therapy for chronic pain is very limited but suggests an increased risk of serious harms that appears to be dose-dependent. More research is needed to understand long-term benefits, risk of abuse and related outcomes, and effectiveness of different opioid prescribing methods and risk mitigation strategies.