Telephone interviewing is increasingly being used to obtain data on health issues. Propertly applied telephone interviewing may have considerable cost benefits, but careful thought has to be put into the design of surveys, weighting and analysis of data to avoid major sources of bias. This study is a hypothetical exercise comparing health estimates from a systematic, self-weighting, multistage, clustered, area sample of households using a face-to-face interview method, with hypothetical samples of people obtained from Random Digit Dialling and Electronic White Pages. In a comparison of the population health estimates obtained for a number of health problems in a hypothetical analysis of these samples, the confidence intervals for the estimates overlapped. Since the estimates are not statistically significantly different, it appears that well-planned, appropriately weighted and analysed telephone surveys can be a less expensive way of obtaining health information, however, some caution is expressed in using this method.