Because of fears that Polaroid colour prints produced with a non-mydriatic fundus camera may not detect important sight threatening lesions in diabetes a study was conducted comparing retinal images obtained on Polaroid prints taken in "field" conditions with those on 35 mm transparencies and fluorescein angiograms. Almost one in five (22/127) Polaroid prints could not be assessed owing to poor quality compared with 3 (2.4%) 35 mm transparencies and 2 (1.6%) fluorescein angiograms. The pick up rate of microaneurysms, haemorrhages, and hard and soft (cotton wool spots) exudates was equivalent for Polaroid prints and 35 mm transparencies of equivalent quality. In two cases with disc new vessels, however, these were not seen on the Polaroid prints. The widespread use of Polaroid colour prints obtained with a non-mydriatic camera without the necessary operative and interpretive skills further limits the usefulness of the technique.