OBJECTIVE—A prospective study was undertaken on the diagnosis of herpes simplex encephalitis (HSVE) by comparing chemiluminescence assay (CL) and two different sensitive polymerase chain reactions (PCRs). METHODS—The materials comprised 53 serial CSF samples from 31 patients with acute encephalitis with suspected HSVE. Each CSF was distributed to three independent laboratories to perform quantitative measurements by CL, the low sensitive (single) PCR, and high sensitive (nested) PCR. The CL provided a method of detecting HSV itself and the small fragment with HSV antigenicity which was composed of viral component proteins. The serial CSFs were found retrospectively to comprise 24 samples from 11 patients with HSVE due to HSV1 and 29 samples from 20 patients with non-HSVE. RESULTS— the CL showed 50 to 48 000 pfu/ml in all samples of HSVE (except one) taken from the 3rd to the 25th day. The low sensitive PCR demonstrated 50 to 47 000 pfu/ml in only six samples of HSVE. The high sensitive PCR disclosed less than 100 to 120 000 copies/ml in 11 samples of HSVE. At the acute stage from the 1st to 7th day, the sensitivities of CL and the high sensitive PCR were 100%, but that of the low sensitive PCR was 75%. The sensitivity of CL was significantly higher than those of both PCRs after the acute stage on the 15th to 32nd day. The specificities and positive predictive values of the three methods were 100%. However, the negative predictive value of CL was significantly higher than that of the low sensitive PCR. CONCLUSIONS—The sensitivity of CL is equivalent to that of the high sensitive PCR during the acute stage and significantly higher than that of the high sensitive PCR after the acute stage. A clear difference in sensitivity exists between the different PCRs. A combination of the PCR, chemiluminescence assay, and serological antibody diagnosis is currently considered the most effective approach for the clinical diagnosis of HSVE.