Aims: To describe the use of a validated diabetes register for sampling frame generation and assessment of the representative nature of participants in a field-work study of diabetic eye disease. Methods: We performed an observational, cross-sectional fieldwork study of diabetic retinal disease using reference standard eye examination. We sampled the entire diabetic population using the Diabetes Audit and Research in Tayside Study (DARTS) diabetes register. Results: The study population comprised 4825 diabetic patients aged over 16 years registered with one of 166 general practitioners (GPs) in 41 practices in Tayside in October 1999. This represented 61.1% of the Tayside diabetic population (7903). A total of 586 (66%; 95% confidence interval 63, 70) patients were examined from a sampling frame of 882 living patients registered with a Tayside GP. Demographic and disease parameters recorded on the DARTS patient register allowed comparison between participants and non-participants. Conclusions: This study shows the clear benefit of using a complete diabetic population as a sampling frame. This allows potential selection bias and external validity to be evaluated using routine data sources. Studies performed and reported in this way will aid the critical appraisal process.