Isolated human populations represent good candidates for studying genetic and environmental causes of common complex diseases because of their decreased genetic and environmental diversity. The possibility of inexpensive and reliable detection of disease prevalence in such populations is therefore of considerable importance, as comprehensive routine health data and disease registries are rarely available in these populations. In this study, we validated the performance of the WHO Rose Angina Questionnaire (RQ) in measuring the burden of coronary heart disease (CHD) in 9 settlements in these Croatian Adriatic islands. CHD was defined as myocardial infarction (MI) diagnosed by a specialist in the local general hospital, or angina pectoris (AP) by a local general practitioner (GP). The true prevalence of CHD in 1,001 adult persons was 10.5%. The results of the RQ screening based on the first 3, 5 and 6 questions were compared with medical record of CHD. Increasing the number of RQ questions from 3 to 6 resulted in decreasing test sensitivity (from 59.0% to 30.5%) and increasing test specificity (from 86.3% to 93.0%) in the prediction of true CHD status. CHD prevalence was overestimated by 76% when subset of the first 3 questions of RQ was used and by 25% when the first 5 questions were used. However, it was underestimated by 10% when the first 6 questions were used. We conclude that RQ is a useful screening method for measuring burden of CHD in isolate human populations, and that the result based on the first 6 questions is a good approximation of the true CHD prevalence in the population, although it should be considered a slight underestimate.