Objectives. Knowledge of infection is essential for human immunodefici
ency virus-type 1 (HIV-1) treatment initiation and epidemic control. T
his study evaluates infection knowledge among infected injection drug
users and acceptance of confidential testing among injection drug user
s, particularly those infected with HIV-1. Methods. A total of 810 inj
ection drug users entering treatment in Contra Costa County, Calif, we
re examined. Clients were tested with unlinked (blinded) tests and sim
ultaneously counseled and offered voluntary confidential HIV-1 antibod
y testing. Data on confidential testing acceptance, previous testing,
drug use, and demographic information were collected. Results. Of the
810 tested, 105 (13.0%) were infected. The current confidential test w
as accepted by 507 (62.6%). HIV seroprevalence in the unlinked survey
was four times greater than in the voluntary survey (13% and 3.5%, res
pectively). HIV-1 infection was associated with refusal of a confident
ial test largely because most infected injection drug users (n = 58; 5
5.2%) already knew of their infection. Of the 47 injection drug users
who were not aware of their infection, 12 (25.5%) accepted the test. A
lthough African-American injection drug users presented with a higher
infection rate (37.3%), they were three times less likely to know of t
heir infection. Conclusions. ''In-clinic'' HIV-1 testing is highly acc
epted, and most infected clients in treatment will learn their status.
Nevertheless, voluntary testing data are likely to yield considerable
underestimates of the true rate of infection among injection drug use
rs.