To quantify the extent that South Carolina HIV/AIDS cases could have been diagnosed during a prior arrest we designed a retrospective population-based cohort study linking case reports from HIV/AIDS Reporting System (HARS) and the South Carolina Law Enforcement Division database. Data from individuals newly diagnosed between January 2001 and December 2005 were linked with statewide arrest records from April 1991 through November 2005. Criminal history data for this report were derived from 28 state prisons and more than 250 law enforcement agencies (jails, lockups, detention centers). Odds ratios and 95% confidence intervals were used to determine how demographic variables and arrest reasons affects receipt of HIV testing. There were 1961/4036 (48.6%) incident cases of HIV diagnosis that had at least one arrest prior to their first positive HIV test. When restricted to 1286/1961 (65.6%) individuals most likely to have been HIV-infected at the time of arrest, 592 (46%) were early testers (no AIDS within 1 year) and 694 (54%) developed AIDS more than 1 year of testing (late testers). After controlling for gender, age, race, behavioral risk and source of HIV report, the odds of being a late tester increased with age (p < 0.001). Overall, 3750 separate arrests were recorded for these 1286 individuals and 491 (13%) arrests were for drug and alcohol or sex crimes. Individuals with 4 or more arrests were more likely to be late testers when compared to those with fewer than 4 arrests (adjusted odds ratio [AOR] 3.30; 95% confidence [CI] 2.28, 4.72). Correctional facilities present considerable opportunities to identify individuals with undiagnosed HIV infection. Providing correctional facilities with the infrastructure for implementation of routine HIV testing would consequently have a significant impact on the health status of the entire community.