The number of plaque forming units (p.f.u.) of virus dispensed on to interferon-treated monolayers in a plaque reduction assay has been assumed to follow a Poisson distribution with the mean estimated from plaque counts on untreated monolayers. From these considerations, plaque counts have been assigned weights and the probit regression line calculated by the method of maximum likelihood. A chisquare analysis of plaque reduction data indicates agreement with the calculated line and does not refute the assumptions leading to the use of Poisson weighting coefficients or a log-normal distribution of p.f.u. ‘tolerances’ to interferon. This implies that the most precise estimate of dose occurs where plaques develop from 27% of the added p.f.u. The calculation of a weighted probit regression line from interferon plaque reduction data has several advantages: all the data from an experiment can be properly utilized; the variance and confidence interval of the interferon titres can be obtained; a direct comparison of dose-responses and titres can be made between different samples; a measure of the reliability of an experiment can be obtained by chi-square analysis.