It has been suggested that the appropriate length of insertion for double-lumen tubes can be estimated by external measurement. This study examined the accuracy of external measurement in estimating the actual length of insertion required in 130 patients. It also examined the relationship between the length inserted and the patient's height in 126 patients and their weight in 125 patients. Although there was a fair correlation between the measured external length and the final inserted length (r = 0.61), the 95% confidence intervals of slope and intercept allowed a large variation and the prediction was too wide to be clinically useful. Height was reasonably well correlated with the final length (r = 0.51) but an equally wide 95% confidence interval rendered it of little clinical value. There was no correlation between weight and final tube length. It is concluded that external measurement alone is not adequate to predict a clinically acceptable position of the double-lumen tube.