BACKGROUND: We are confronted with a number of extremely contradictory results concerning the spontaneous remission of addicts. This ranges from very low remission rates in treated addicts to extremely high remission rates within short periods of time in untreated addicts ***METHODS: In order to shed light on these contradictions, some well-known publications were analysed, and theoretical reasons were sought as to why these contradictions occur. ***RESULTS: There are substantial artifacts, easily explained in probabilistic terms, explaining why remission rates (whether therapy-induced or spontaneous) are systematically underestimated in clinical contexts and systematically overestimated in epidemiological studies. The systematic underestimation is primarily explained by a form of Berkson's artifact that Cohen and Cohen (1984) labelled this “The clinician’s illusion”. The overestimation is primarily explained by a regression to the mean artifact (Campbell & Kenny, 1999). Both artifacts have been known for many decades but are usually neglected when discussing results. In addition, there are sometimes highly questionable assumptions made when interpreting data that were collected on a routine basis (e.g. Winick, 1962 or Robins et al., 1974). ***CONCLUSION: The above problems should be given much greater focus in empirical studies focusing on assessing remission. Some biases can be mitigated or prevented when designing the study, and some can be partially or fully compensated when analysing the data. For existing studies showing enormous remission rates, it is usually not possible to assess the extent to which these reflect real changes and the extent to which they reflect artifacts.