Borderline personality disorder (BPD) is characterized by a pervasive pattern of instability and impulsivity. Several North American prospective studies support the high level of mental health care utilization in this population. There is little data in other systems of health organization, such as France. Furthermore, little is known on the variables associated with the mental health service utilization among BPD patients. The main objective was to compare the utilization of mental health care among BPD patients, to the general population and patients with another personality disorder (PD) and to describe the demographic and clinical factors associated with the group of patients who use the most health care. A multi-center (5 public and private centers), epidemiological study. Data were collected prospectively (database of an insurance fund covering 80% of the population) and viewed, retrospectively. We used the data collected during the five years previously to the inclusion. Inclusion criteria were age (18-60 years) and membership in the health insurance fund targeted. Patients on legal protection, forced hospitalization, with a chronic psychotic disorder, manic, mental retardation, or not reading French were excluded. First, four groups were composed: BPD, other PD, control groups for PD and other PD. The first two groups were recruited from a screening of inpatients including a self-administered questionnaire (Personality Disorder Questionnaire 4+). Assessment by a psychologist including the Structured Interview for DSM-IV Personality Disorders (SIDP-IV) was given straight to those who had a score above 28. This questionnaire allowed us to distinguish one group of subjects with BPD and a group with other PD (without BPD). Clinical evaluation included Axis I (MINI), Axis II (SIDP-IV), psychopathological features (YSQ-I, DSQ-40), demographic variables and therapeutic alliance (Haq-II). Matched controls (age, sex) composed the 3rd and 4th group (BPD control and other PD control). They were randomly chosen in the health database insurance previously used. One hundred and thirty-seven (95.8%) screened patients agreed to answer the psychological assessment. In this sample, 44 (32.1%) had BPD, 39 (28.5%) other PD and another 39 (28.5%) did not have PD. The BPD group was compared to a sample of 165 matched subjects and the other group PD to a sample of 123 matched controls. There was no difference between BPD and other PD groups regarding the mental health utilization. However, there was an increased use of hospitalizations and deliverances of nervous system drugs in both clinical groups compared to their controls. The analysis of drugs supplied in pharmacies for BPD patients showed that the first two drugs were opiate substitutes (12.3% methadone, buprenorphine 6.7%). No anticonvulsants or atypical antipsychotics appear in the top 20 of treatments delivered. A composite variable (hospitalization for more than 6 months during previous five years and 500 supplied drugs) allowed the discrimination of two groups among patients with BPD: heavy users of care and low care users. No variables (demographics, Axis I, Axis II, self-aggressiveness, DSQ-40, Haq-II, YSQ-I) could discriminate the two groups except the number of previous psychotherapies (heavy users: n=0.4 (SD 0.5) vs low users: n=1.8 (SD 2.1) P=0.0054). This study confirms the important use of the service of BPD patients in France, as well as the possible moderating role of psychotherapy. We found a mismatch between these uses and recommendations.