Opioid dependence often is a chronic relapsing disorder. The evidence for medication-assisted treatment (MAT) being the treatment of first choice is unequivocal. Yet, health systems, professionals and patients often fail to offer or, respectively, to enter long-term and low-threshold MAT and instead treat opioid dependence as a semi-acute disorder. We claim that the typical perspective on treatment retention, one of the pivotal outcomes used for evaluation of MAT, contributes to this phenomenon by obscuring the chronic nature of opioid dependence. To support this hypothesis, we use data of 11,819 patients obtained from the MAT register of the canton of Zurich 1991-2015 and illustrate three different perspectives on treatment retention: the prospective "trial" perspective looking at the duration of single treatment episodes; the retrospective "provider" perspective looking at the duration of treatment episodes of patients still in treatment; and the "public health" perspective looking at the proportion of days spent in treatment regardless of single treatment episodes. Integrating these perspectives will lead to a more realistic perception of the chronic nature of opioid dependence. This will allow determining more appropriate aims and time frames for MAT in practice and research.