OBJECTIVE: To estimate the prevalence of drug use disorders in the European Union and Norway. METHOD: Based on a systematic literature search and an expert survey, publications after 1990 on prevalence of drug use disorders (DUD, defined as drug dependence and drug abuse or harmful use) in EU countries and Norway were reviewed. The search included both direct estimations based on general population surveys using the DSM-IIIR, DSM-IV, or ICD-10 definitions of DUD; and indirect estimates based on other epidemiological methods, such as multiplier procedures based on treatment or legal data and capture-recapture techniques. The indirect methods did not use diagnostic criteria, but criteria based on duration and pattern of use, labelled as problematic drug use as a meaningful approximation. RESULTS: The majority of DUD as estimated from direct methods using general population surveys were cannabis use disorders, usually not included in indirect estimates. The prevalence of thus defined disorders can be as high as 3%. For disorders other than cannabis use disorders (i.e. opioid, cocaine and amphetamine use disorders), indirect estimates of prevalence were consistently higher than those based on direct estimates, and ranged between 0.3% and 0.9% in European Union countries and Norway. Men have higher prevalence rates of DUD than women, but the difference was much less pronounced in general population surveys. Younger age (18-25 years) is the age group with the highest estimates. CONCLUSION: General population surveys typically result in a serious underestimation of the prevalence of DUD other than cannabis use disorders, because many people with DUD are not reached by these surveys (hidden populations). Based on the more valid indirect estimates, it is concluded that problem drug use constitutes a relatively high burden of disease and social problems in Europe.