The aim of this study is to evaluate the concordance between predicted postoperative forced expiratory volumes in 1s (ppoFEV1) calculated on the basis of data known before surgery with ppoFEV1 calculated after completing surgical procedure. We have prospectively studied 66 consecutive patients (55 cases scheduled for lobectomy and 11 for pneumonectomy) operated on for bronchial carcinoma. According to location, 33 tumours were classified as central and 33 as peripheral. In all cases, ppoFEV1 was calculated twice: first (ppoFEV1-A) according to the scheduled surgical procedure; second (ppoFEV1-B) according to the procedure eventually performed. At operation, 43 lobectomies (65.2%) and 23 pneumonectomies (34.8%) were performed. Differences between ppoFEV1 A and B were found in 18 cases (12 central tumours). In three of them (4.5% of 66 cases), ppoFEV1-B was under 40%. Pearson coefficient was 0.85 (P<0.001) for the whole series of cases; 0.83 (P<0.001) for central and 0.87 (P<0.001) for peripheral tumours. On multiple regression analysis, R2 was 0.76 and ppoFEV1-A had the highest influence on the dependent variable. We have found that: (1) there is no perfect correlation between ppoFEV1 calculated with data known before and after surgery; (2) discrepancies are most important in centrally located tumours and (3) in 4.5% of cases, discrepancies could have influenced the preoperative risk estimation.