Fifty-three patients from a general practice and 33 subjects who were in a community-screening promotion underwent total cholesterol level measurements both with a portable Reflotron system and by a standardized laboratory procedure. The intraclass correlation coefficient between the paired measurements was equal to 0.956; this suggests that the results of the Reflotron system agreed very closely with those of the laboratory procedure. In spite of this agreement, the Reflotron instrument, on average, gave lower results than did the laboratory procedure (mean difference, -0.164 mmol/L; 95% confidence interval, -0.094 to -0.234 mmol/L). This difference is small and, in comparison with other sources of variation in cholesterol measurement, is unlikely to be important in day-to-day clinical practice. However, the difference between the two methods appeared to be related to the cholesterol level (larger differences at lower cholesterol levels). In large studies, where small differences in cholesterol levels are important, this difference should be considered against the obvious convenience of rapid portable cholesterol measurements.