OBJECTIVE: To compare the Mallampati classification and Wilson risk-sum and adopt a predictive rule in our setup that has better results. DESIGN: An analytical study. PLACE AND DURATION OF STUDY: The main Operation Theatre of Combined Military Hospital, Rawalpindi from 1st September to 31st December 2000. PATIENTS AND METHODS: Three hundred and thirty-eight patients were evaluated pre-operatively for difficult intubation using both the tests. The sensitivities, specificities and positive predictive values (PPV) were determined in grading the laryngeal view in each case during direct laryngoscopy. RESULTS: Both tests identified only 3 out of 7 difficult intubations, giving a similar sensitivity of 0.42. Twice as many patients were predicted to be difficult by Mallampati classification than by Wilson risk-sum (specificity 84% and 93%). The Wilson risk-sum had better positive predictive value (11%) as compared to 5% of Mallampati classification. CONCLUSION: The Wilson risk-sum had preferred for assessment of the airway because of its better specificity and positive predictive value while noting that both tests have poor sensitivities when used alone.