Objectives: To evaluate the process and outcomes of HIV partner notification (PN) activity in Scotland. Design: Retrospective population based study. Subjects: 114 adults newly diagnosed with HIV infection (index patients) in Scotland between September 1995 and August 1996. Setting: Healthcare settings in which all 114 new HIV diagnoses were made: 42 (37%) from genitourinary medicine; 32 (28%) infectious diseases; 18 (16%) general practice; and 22 (19%) from other sites. Main outcome measures: Number of partners notified and tested up to 9 months after initial diagnosis. Results: Of 114 index patients (IPs), information on current partners was available for 102 (89%). PN was not appropriate for 47 of the 102 IPs. The remaining 55 IPs identified 63 current partners at risk, of whom 51 were notified: 44 underwent HIV testing, which yielded 11 new HIV positive diagnoses. Information on previous partners was available for only 56 IPs (49%). PN was not appropriate for 30 of the 56 IPs; the remaining 26 IPs identified 46 previous partners at risk, of whom 12 were notified: four were tested, but yielded no new diagnoses. Conclusions: Notification of current partners was performed well and was an effective strategy for identification of HIV positive individuals at a presymptomatic stage. Notification of previous partners was limited. Partner notification was attempted in a wide range of healthcare settings. Given the clinical effectiveness of antiretroviral therapy, partner notification as a tool towards early diagnosis of HIV disease deserves renewed attention.