Auscultatory percussion is a new method of physical examination developed by Guarino [Lancet i: 1332-1334, 1980]. It consists in tapping lightly the manubrium sterni with the distal phalanx of the middle finger while listening over the chest wall posteriorly with a stethoscope; a decrease in sound intensity is usually attributed to lung abnormalities. The aim of our study was to assess the validity of the method as compared to classical percussion, a point not entirely clear in the original study. Two observers independently examined 281 unselected patients (170 men; 111 women) referred for chest X-ray studies. Roentgenographic analysis, carried out by a third observer, revealed 12 categories of abnormalities in 96 patients. The validity of auscultatory and conventional percussion was calculated by taking the product of sensitivity and specificity of each method. For both observers and both methods, the values of this index were always found to be below 0.25, indicating that the results can be explained entirely by chance. When the roentgenographic category was taken into account, both methods of percussion were valid only to detect large pleural effusions. Auscultatory percussion failed completely to detect many other abnormalities including solitary nodules less than 6 cm in diameter. There is no single hypothesis to explain these results. Various possibilities are discussed.