Acyclovir levels to date have been performed on plasma rather than serum samples. To determine if serum was an acceptable specimen, acyclovir levels were measured on simultaneously collected plasma and serum samples by radioimmunoassay. Eighty-two paired samples from 20 patients containing between 0.39 and 115.3 μM acyclovir were studied. Least squares regression, orthogonal regression, and two-way analysis of variance with replication all indicated an acceptably high degree of correlation. We concluded that serum is a suitable specimen for measuring acyclovir levels.