Background. For estimating the migraine prevalence is very important to know the discrimative property of a scale. Michel's Standardized Diagnostic Questionnaire is a useful tool for this proposal. However, it is possible to improve its diagnostic capacity after adjusting overall punctuation by item analysis. Materials and methods. One-hundred and seventy students of psychology selected at ramdon filled out the Michel's Questionnaire, and after a neurologist made an interview in order to identify a migraine headache. Kraemer's method was used for reassigning a new score taking into account the association of a satured binomial regression model. Area under receptor-operator curve (ROC) was calcuted for the original and new score, comparing by likelihood methods. Results. For the original punctuation, the area under ROC was 0.853, and 13 was the best discriminatory score with sensitivity of 58.6% (95%CI 44.9%-71.4%), specificity 89.3% (95%CI 82.0%-94.3%), and concordance of 78.8% (95%CI 91.9%-84.7%). For the new way of punctuation, the area under ROC was 0.907 (95%CI), the best discriminatory score 18, with sensitivity in 65.5% (95%CI 51.9%-77.5%), specificity 92.0% (85.3%-93.3%), and concordance of 84.7% (95%CI 76.4%-89.3%). The area under ROC from the new punctuation was better than original one (p=0.016). Conclusions. This reformulation of the Michel's questionnaire improves the discriminatory capacity of this scale.