Direct immunofluorescence (IF) staining was compared with virus isolation for detection of herpes simplex viruses (HSV) and varicella-zoster virus (VZV) directly in clinical materials. These included 199 vesicular lesion specimens and 280 tissue specimens. Correspondence between IF and isolation results was 88% in testing for HSV in lesion specimens and 98% in testing for HSV in various tissue (mostly brain) specimens. Overall, IF was positive for 82% of the specimens in which HSV was demonstrated, and virus was isolated from 89% of the HSV-positive specimens. IF was markedly more sensitive than isolation for demonstrating VZV in lesion and tissue specimens, detecting all of the specimens positive for VZV, whereas isolation detected only 23%. IF detected VZV antigen in a number of lesion specimens taken late after onset, past the time when they would be expected to yield infectious virus. Specificity of positive IF reactions for HSV or VZV in the absence of virus isolation was supported by the facts that (i) staining was obtained with only a single, presumably homologous, immune conjugate, (ii) clinical symptoms were compatible with infection with the virus for which positive IF findings were obtained, and (iii) positive electron microscopy findings for herpesviruses or positive serological results for VZV were also obtained in some instances. Factors to be considered in achieving specificity of IF staining for these human herpesviruses are discussed.