The present study investigated the reliability of Sexually Violent Predator (SVP) civil commitment criteria under Florida's Jimmy Ryce Act. The purpose of the study was to determine if, independently, 2 evaluators would reach the same conclusions about the same client (n = 295). According to civil commitment criteria outlined by the United States Supreme Court (Kansas v. Hendricks, 1997), SVPs must display a mental abnormality predisposing them to sexual violence and a likelihood of future sexual violence. The interrater reliability of 8 DSM-IV diagnoses applied by evaluators to determine whether a client has a "mental abnormality that predisposes him to sexual violence" was found to be poor to fair (kappa = .23 to . 70). The interrater reliability of risk assessment instruments used to determine "likelihood of reoffense" was good (ICC = .77 to .85). The recommendations made by evaluators regarding whether or not to refer a client for civil commitment demonstrated poor reliability (kappa = .54). Implications for practice and policy are explored.