BACKGROUND The Chemcard cholesterol test has been designed to estimate cholesterol levels in patients in General Practitioners' surgeries. The patients displaying elevated cholesterol levels may then be referred for a more accurate test at a recognised laboratory. OBJECTIVE To compare the accuracy of the Chemcard cholesterol procedure with a standardised laboratory procedure for estimating cholesterol levels. METHODS The Chemcard cholesterol procedure was applied to 200 subjects from the community, who were enrolled in a cardiovascular risk assessment programme. RESULTS The correlation coefficient between the paired measurements was 0.83. The Chemcard cholesterol procedure gave slightly higher results than did the laboratory procedure as shown by the mean (+/-standard deviation) difference between the two methods of 0.16 +/- 0.58 mmol/L. CONCLUSIONS The poor agreement between the two methods and the potential misclassification by Chemcard suggests that the Chemcard is not a reliable test system. The high level of risk group misclassification found in this study refutes the manufacturer's claims and does not eliminate the need for formal laboratory cholesterol measurement.