This prospective study aimed to evaluate in 849 patients the Mallampati and Wilson scores for predicting a difficult intubation. All scheduled patients were included. Induction and tracheal intubation were carried out as usual. Intubation was deemed to have been difficult if any special procedure had been required (external compression excepted); difficult laryngoscopy was defined as grade 3 or 4 on the Cormack-Lehane scale. In accordance with these criteria, less than 36% of patients with difficult intubation or laryngoscopy were detected, with a high false positive rate (more than 75%) but a good negative predictive value (more than 90%). This study was not concordant with the results of the original studies. However, data from literature show a great variability of results between studies. Many factors may contribute to this variability: differences between samples of patients, evaluation of Mallampati or Wilson scale, protocols of induction and intubation or characterization of difficult intubation. This study suggests a poor reliability of the two tests.