Heroin users in opioid agonist treatment (OAT) show markedly reduced heroin consumption, less crime and a lower mortality rate. However, the extent of long-term OAT participation over subsequent treatment episodes remains unclear. We analysed the annual proportion of patients in treatment (at least 1 day) since the start of first OAT in 4 European regions: Barcelona (BA) 1996–2012: 8,602 patients; Czech Republic (CZ) 2000–2014: 4,377 patients; Netherlands (NL) 1994–2014: 33,235 patients, Zurich (ZU) 1992–2015: 11,795. We estimated the long-term decline of treatment need due to mortality or abstinence and also a “nuisance” short-term decline until the equilibrium level of cycling in and out of OAT is reached to obtain the adjusted treatment participation value. The adjusted treatment participation was around 50% (BA: 47.4–51.4%; CZ: 49.8–53.9%; NL: 52.3–54.0%; ZU: 46.4–49.3%), and the annual decline of treatment need was close to 4%. Non-nationals (NL patients with a migrant background) showed substantial lower adjusted treatment participation (rate ratio BA: 0.059–0.343; NL: 0.710–0.751; ZU: 0.681–0.797; CZ: n.a.). Our method may provide a policy-relevant indicator of long-term provision, quality and access to OAT following first treatment.