The prevalence of HIV infection in correctional settings is several-fold higher than found in community settings. New approaches to identifying HIV infection among prisoners are urgently needed. In order to determine the HIV seroprevalence and to identify the correlates of HIV infection among female prisoners, an anonymous, but linked HIV serosurvey was conducted at Connecticut's sole correctional facility for women (census=1,100). After removing all individual identifiers for inmates' standardized clinical and risk behavior information, data are linked by a third source to blinded HIV-testing information by a third party. This three-step sequential process allows for anonymous HIV testing that can still be linked with deidentified clinical and behavioral data. Of the 3,315 subjects with complete information, 250 (7.5%) were HIV+. Of these, 157 (63%) self-reported being HIV+. Using multiple logistic regression analysis, having sex with a known HIV+ person [adjusted odds ratio (AOR)=9.1] and injection drug use (AOR=6.1) were the most highly correlated risk factors for HIV, whereas leukopenia (AOR=9.4) and hypoalbuminemia (AOR=7.2) were the most significant laboratory markers. Other independent correlates of HIV included self-report of syphilis (AOR=1.9) or genital herpes infection (AOR=2.7) and being Black (AOR=2.1) or Hispanic (AOR=2.2). The prevalence of HIV and HIV-risk behaviors is high among incarcerated women. Existing voluntary HIV counseling and testing programs do not completely target high-risk groups who remain part of the evolving epidemic. Defined demographic, behavioral, and clinical assessments may provide useful information for encouraging targeted counseling and testing. Newer targeted approaches merit further study to determine the effectiveness of this approach. Alternative methods of facilitating more widespread HIV testing, such as saliva tests, rapid serologic tests, and more routine testing in high HIV-prevalence areas should be considered both for clinical and for public health benefits.