Improvement of serologic assays for detection of antibodies against human herpesvirus 8 (HHV-8) is critical to better understand its epidemiology and biology. We produced the HHV-8 latent (ORF73) and lytic (ORF65, K8.1, and glycoprotein B) antigens in the Semliki Forest virus system and evaluated their performance in immunofluorescence assays (IFAs) and enzyme-linked immunosorbent assays (ELISAs). These assays were compared with other latent antigen-based assays, including an IFA based on primary effusion lymphoma (PEL) cells and an ELISA based on bacterially expressed ORF73 antigen, as well as with other lytic antigen-based assays, including an IFA based on induced PEL cells, a commercial ELISA based on purified virions, and ELISAs based on K8.1- and ORF65-derived oligopeptides. We used a panel of 180 serum specimens obtained from three groups expected to have high, intermediate, and low HHV-8 prevalences. Using three different evaluation methods, we found that (i) the performances of the lytic antigen-based ELISAs were almost equivalent, (ii) the lytic antigen-based assays were more sensitive than the latent antigen-based assays, and (iii) in general, IFAs were more sensitive than ELISAs based on the same open reading frame. We also found that serum specimens from healthy individuals contained antibodies cross-reactive with HHV-8 glycoprotein B that can potentially cause false-positive reactions in lytic PEL-based IFAs. Although this is not a substantial problem in most epidemiologic studies, it may confound the interpretation of data in studies that require high assay specificity. Because the K8.1-based IFA provides sensitivity similar to that of lytic PEL-based IFAs and improved specificity, it can be a useful alternative to the PEL-based IFAs.