Data for two states have been extracted from the Highway Safety Information System and used to examine the effect of median width on the frequency and severity of accidents. Log-linear models for accident rates have been used to describe the effect of median width after adjusting for other variables. Effects have been estimated by the quasi-likelihood technique assuming a negative binomial variance for the accident count per roadway section. Results for both states indicate that total accident rates and rates for specific types and severity decline rapidly when median width exceeds about 25 ft (7.6 m). Policy guidelines for median widths are somewhat nebulous, partly due to the lack of large well conducted studies providing quantitative information on this topic. The results provide a basis for the development of more precise guidelines regarding median width.