Various methods are in use for estimating physiological variables from maximum expiratory flow-volume curves, and some of these are recommended by international bodies for use on adults. We have compared eight such methods in 433 adolescents aged 12-16 years. The mean values of expiratory flows, FVC and FEV1 by each method differed less than 6% from those obtained by the method recommended by the ATS, except for MEF25, where differences between methods were substantially larger. The mean within-individual coefficients of variation of FVC and FEV1 did not differ between methods and improved from about 3.1 to 2.6% and 2.7 to 2.3% respectively when using 5 instead of 3 acceptable curves. For expiratory flows, the coefficients of variation ranged between 4.2 and 9.4% and improved slightly but not consistently when using more than 3 acceptable curves. Reproducibilities were systematically poorer when flows were derived from one "best" FVC manoeuvre. It is concluded that taking more than 3 acceptable blows is not cost-effective, even in inexperienced adolescents. An envelope method and maximum composite curve from FVC reproducible within 5% from the largest one are recommended for use with teenagers.