The WHO/Rose Questionnaire on intermittent claudication was developed in 1962 for use in epidemiological surveys, and has been widely used. Several population studies have shown, however, that it is only moderately sensitive (60-68%), although highly specific (90-100%). In this study, reasons for the poor sensitivity and good specificity were determined following its application to 586 claudicants and to 61 subjects with other causes of leg pain. The results showed two important findings: firstly, that over half of the false negatives were produced by one question alone; and secondly that only three questions were required to maintain the specificity of the questionnaire. This knowledge, in conjunction with the pre-testing of additional questions, enabled a new questionnaire to be constructed: the "Edinburgh Claudication Questionnaire". This questionnaire was tested on 300 subjects aged over 55 years attending their general practitioner, and found to be 91.3% (95% CI 88.1-94.5%) sensitive and 99.3% (95% CI 98.9-100%) specific in comparison to the diagnosis of intermittent claudication made by a physician. The repeatability of the questionnaire after 6 months was excellent (kappa = 0.76, p < 0.001). These results suggest that this revised version of the WHO/Rose Questionnaire should be adopted for use in future epidemiological surveys of peripheral vascular disease.