134 swabs in viral transport medium were received from 126 patients with suspected clinical HSV-1 and HSV-2 infections. They were tested by (i) nested multiplex polymerase chain reaction NMPCR (strongly positive specimens had visible bands on both rounds of PCR) without prior extraction, (ii) culture in primary rhesus monkey kidney, E6-Vero, RD and HEp-2 cells and (iii) antigen detection by immunofluorescence (IF). Antigen detection employed four novel pools (A-D) of monoclonal antibodies (Mab): A was HSV-1 specific, B was HSV-2 specific while C and D were generic. In comparison to NMPCR the sensitivity and specificity of (i) culture was 59% (22/37) and 100% (134/134), (ii) IF by Pool A was 59% (16/27) and 100% (117/117), (iii) IF by Pool B was 40% (4/10) and 100% (130/130) and (iv) IF by Pools C and D were 60% (18/30) and 100% (96/96). Specimens positive by culture were more likely to be strongly positive by NMPCR (chi2 P = 0.004). Typing by each method concurred on all occasions. NMPCR was cost effective, easier to perform and was the most sensitive method for HSV detection. It should become the method of choice for HSV diagnosis.