In literature, there exists disagreement regarding the impact of comorbid personality disorder(s) (PD[s]) on treatment outcome for patients with major depressive disorder (MDD). The aim of this paper was to investigate whether statistical heterogeneity and inter-dependency are potential moderators of the effect of co-morbid PDs on outcome for patients with MDD. Clinician-rated MADRS scores and DSM-IV Axis II personality disorder diagnoses were obtained from 562 outpatients with MDD who received 6 months of combined psycho- and pharmacotherapy. Single-level regression showed significantly worse treatment outcome among patients with co-morbid PD, as compared to patients with no PD. After controlling for statistical heterogeneity and interdependency, treatment outcome was no longer significantly worse for patients with co-morbid PD. In conclusion, heteroscedasticity and inter-dependency should be considered as potentially compelling explanations for inconsistencies in findings on treatment outcome for depressed patients with co-morbid PDs.