Little is known about the mechanisms through which routine outcome monitoring (ROM) influences psychotherapy outcomes. In this secondary analysis of data from a randomized clinical trial (Brattland et al., 2018), we investigated whether the working alliance mediated the effect of the Partners for Change Outcome Monitoring System (PCOMS), a ROM system that provides session-by-session feedback on clients' well-being and the alliance. Adult individuals (N = 170) referred for hospital-based outpatient mental health treatment were randomized to individual psychotherapy either with the PCOMS ROM system, or without (treatment as usual [TAU]). Treatment was provided by the same therapists (N = 20) in both conditions. A multilevel mediation model was developed to test if there was a significant indirect effect of ROM on client impairment at posttreatment through the alliance at 2 months' treatment controlled for first-session alliance. Alliance ratings increased more from session 1 to 2 months' treatment in the ROM than TAU condition, and alliance increase was associated with less posttreatment impairment. A significant indirect effect of ROM on treatment outcomes through alliance increase (p = .043) explained an estimated 23.0% of the effect of ROM on outcomes. The results were consistent with a theory of the alliance as one mechanism through which ROM works. (PsycINFO Database Record (c) 2019 APA, all rights reserved).