An evaluation was made of the use of telephone survey methods to collect illicit drug use data. Using data from a national survey that collects data by personal interviews, marijuana and cocaine use prevalence rates among households with telephones and those without were compared in order to assess coverage errors in telephone surveys. Drug use rates were substantially higher among households without telephones, with 24.9 percent of those living in households without telephones reporting use of marijuana in the past year, compared with only 9.4 percent of persons living in households with telephones. Trends in drug use were divergent, with substantial decreases in use occurring between 1985 and 1988 in households with telephones, but not in those without. National prevalence patterns and trends among households with telephone appear to be consistent with national patterns and trends in the total household population, because about 93 percent of the population lives in households with telephones. However, surveys conducted by telephone were found to produce underestimates of illicit drug use prevalence. In a 1988 national telephone survey, estimated rates of past year use were 5.2 percent for marijuana and 1.4 percent for cocaine. Comparable data from a personal visit survey (including only households with telephones and reedited and reweighted to control for differences in data collection protocols) were 8.0 percent for marijuana and 3.1 percent for cocaine use. Comparisons with several other telephone surveys collecting illicit drug use data showed similar results. Based on these results, researchers are advised to use caution in using telephone surveys to produce drug use prevalence estimates.