Background The stability of personality disorders in substance abusers have been questioned. Previous research has suggested that personality disorder features change when substance use is reduced, but remain stable when substance use is continued. The current study extended this finding to alcohol patients. Materials and Methods Alcohol dependent subjects treated for alcohol problems were tested with the Millon Clinical Multiaxial Inventory II (MCMI-II) at intake to inpatient treatment, discharge from treatment, and at 6 months follow-up. Test-retest correlations were calculated and compared with discriminant correlations, and measures of change were calculated. The association between percent of days abstinent and change in MCMI-II personality disorder scales was assessed using hierarchical regression controlling for baseline scores. Results MCMI-II personality disorder scales were significantly correlated across assessment waves, and comparison violations were relatively few. Large changes occurred during treatment, and only small changes from discharge to follow-up. Changes were associated with percent days abstinent for avoidant, dependent, and masochistic personality disorder scales. No significant differences were observed in the stability between the control and experimental condition. Conclusions The MCMI-II was found to be stable, and little evidence was found that there was a lack of discriminant validity. Changes in personality disorder severity as measured by self-report took place during treatment, rather than after. There was some indication that decreases in personality disorder scales from discharge to follow-up were associated with abstinence from drinking during follow-up.