Background: Treatment integrity is defined as the extent to which a treatment is carried out as intended. It includes several interrelated components (e.g. therapy adherence, therapist competence and therapy differentiation). Treatment integrity is regarded to play an essential role in psychotherapy outcome research. Objective: This article discusses (1) conceptual aspects, (2) empirical results and (3) methodological problems of research on treatment integrity. Results: (1) Therapy integrity refers to different aspects of experimental validity. It is usually discussed with regard to internal validity. Maybe even more important, however, treatment integrity also refers to construct validity. Treatment integrity is also related to external validity and validity of statistical conclusions. (2) Results for the relationship between treatment integrity (adherence to a treatment model and competent delivery of techniques) and outcome are heterogeneous and suggest that the relationship between outcome and treatment integrity is not yet clear. Even in studies which found a relationship between treatment integrity and outcome, the proportion of variance explained by intended techniques was between 10% and 30%. Thus, the majority of variance was not explained by intended techniques. (3) Methodological problems may contribute to the lack of consistent results. As most results come from randomized controlled efficacy studies, the ranges of data for both outcome and adherence/competence may be restricted by the selection of patients and therapists, as well as by training of therapists, manualization and monitoring of treatment. These factors attenuate the association between treatment integrity and outcome. Other factors that may contribute to the inconsistent results are unreliability of measures, instability of adherence and competence during the process of psychotherapy or nonlinear relationships between treatment integrity and outcome. Conclusions: Further studies are required that address the relationship between adherence, competence and outcome, especially for those methods of psychotherapy that were shown to be effective in a particular disorder. Factors that moderate or mediate the relationship between outcome and therapy integrity should be included. Due to their high level of standardization, randomized controlled efficacy studies seem to be only of limited use to study the relationship between treatment integrity and outcome. The recent shift from a single-disorder focus of the traditional manualized interventions to transdiagnostic, modular and component-based interventions may lead to a change in the conceptualization of treatment integrity. It allows for a more flexible use of interventions while maintaining treatment integrity. This shift may both enhance the external validity of studies using manualized treatments and lead to more consistent results regarding the relationship between treatment integrity and outcome.