Perfectionism has been seen to predict treatment outcome and symptom severity in clinically relevant diagnoses such as depression and anxiety disorders. It has also been suggested to be a transdiagnostic phenomenon. The aim of this study was to investigate, for the first time, predictors of change in Internet-based cognitive behaviour therapy for perfectionism. The chosen predictors were perfectionism severity, comorbid diagnosis and degree of self-compassion. The study was a part of the Devin-project; a randomized controlled trial with 150 participants that were allocated to active treatment (N=73) or a waitlist condition (N=77). Multiple linear regression analysis was used for the analyses. Variables chosen a priori did not significantly predict the treatment outcome (R2=.10, p = .099). A post hoc-analysis showed that perfectionism severity, as measured with CPQ, predicted symptom change after treatment (B = .783, β = .56 p < .001). The results indicate that perfectionism severity doesn’t impair the treatment effects and that the treatment is effective, independent of comorbid diagnoses. Self-compassion was investigated on an explorative basis but the results can indicate that the variable is not a risk factor nor a protective factor for perfectionism. As Devin is the largest randomized controlled study of internet-based cognitive behaviour therapy for perfectionism to date, this study generates important implications and suggestions for future research. / Devin