The functioning of an external quality-control scheme for amino acids set up in 1978 is described. Two measurements were made each month by participating laboratories on a control plasma sample provided by the quality-control center; freeze-dried samples were used from 1978 to 1989 and liquid samples since 1990. In addition, two "blind" samples were sent to the laboratories each year. Every 3 months, the overall results and those of the individual laboratories were analyzed statistically. The validity of the liquid sample control is demonstrated. The progressive improvement of results is commendable. In 1990, the coefficients of variation for all participants ranged from 8.4% for alanine to 23.5% for methionine. The standards used for calibration could contribute to the broad range of results, especially those for histidine and ornithine. The use of blind samples made it possible to detect problems of calibration, of linearity of measurement, of contamination, and of identification of unusual amino acids.