Thirteen laboratories participated in blind tests of a panel of 20 coded cerebrospinal fluid specimens (7 uninfected samples, 3 samples infected with 1 50% tissue culture infective dose [TCID50]/0.1 ml [nonenterovirus strains], and 10 samples infected with 10, 1, or 0.1 TCID50/0.1 ml [three different enterovirus serotypes]) on the Amplicor enterovirus PCR assay (Roche Diagnostic Systems). The panel was also evaluated by in-house PCR (two nested-PCR and three one-step PCR assay) or tissue culture (eight laboratories). The viral load was shown to influence greatly the sensitivity of the assay. The average sensitivity of the Amplicor test ranged from 67 to 98% for viral titers of 1 to 10 TCID50/0.1 ml, respectively; titers of 0.1 TCID50/0.1 ml resulted in a sensitivity of only 16%. The overall specificity of the Amplicor test was 98%. The Amplicor assay compared favorably to the five in-house PCR tests (no significant difference in either sensitivity or specificity) and was much more sensitive than tissue culture (P < 0.001), even for high viral loads. It was easy to perform, rapid (about 6 h), well-standardized, and appeared to be suitable for the diagnosis of enterovirus meningitis on a routine basis in laboratories trained in molecular biology techniques.