Objective: Alcohol and Other Drug (AOD) disorders cause substantial harm. Effective Substance Use Treatment (SUT) exists, but long-term outcomes remain inconclusive. This study used a 20-year prospective follow-up of 1248 service users entering SUT in Stockholm, Sweden, in 2000–2002 to elaborate on how different dimensions of long-term outcomes may be measured by register-based indicators. Baseline characteristics and attrition bias were explicated, and register-based outcomes were examined.Results: Register-based indicators are valuable, but they also have inherent limitations such as the lack of substance use data and inability to differentiate between un/met treatment needs and access. Significant variations in long-term outcomes were evident depending on which register-based indicator was used, and whether used in isolation or combinations. Six out of 10 service users were still alive after 20 years, but as many as 8 out of 10 of the survivors remained in treatment, and only two out of 10 had a stable economic situation. Hence, the register indicators identified only a few survivors, with stable economic and social situations, and without recent treatment contacts 20 years after treatment entry. The long-term outcomes were concerning and even more so when combining outcome dimensions.