Objective: The purpose of this study was to assess the lifetime rates of oc
currence of a full range of DSM-III-R axis I disorders in a group of patien
ts with criteria-defined borderline personality disorder and comparison sub
jects with other personality disorders. Method: The axis I comorbidity of 5
04 inpatients with personality disorders was assessed by interviewers who w
ere blind to clinical diagnosis and who used a semistructured research inte
rview of demonstrated reliability. Results: Four new findings emerged from
this study. First, anxiety disorders were found to be almost as common amon
g borderline patients (N=379) as mood disorders but far more discriminating
from axis II comparison subjects (N=125). Second, posttraumatic stress dis
order (PTSD) was found to be a common but not universal comorbid disorder a
mong borderline patients, a finding inconsistent with the view that borderl
ine personality disorder is actually a form of chronic PTSD. Third, male an
d female borderline patients were found to differ in the type of disorder o
f impulse in which they "specialized." More specifically, substance use dis
orders were significantly more common among male borderline patients, while
eating disorders were significantly more common among female borderline pa
tients. Fourth, a lifetime pattern of complex comorbidity (i.e., met DSM-II
I-R criteria for both a disorder of affect and a disorder of impulse at som
e point before the patients' index admission) was found to have strong posi
tive predictive power for the borderline diagnosis as well as a high degree
of sensitivity and specificity. Conclusions: These results suggest that th
e lifetime pattern of axis I comorbidity characteristic of borderline patie
nts and distinguishing for the disorder is a particularly good marker for b
orderline personality disorder.