Mortality predictive factors of burned patients are analyzed in a population of 1929 patients by the statistical method of logistic regression. Among the variables studied (total burn skin area, deep burn area, superficial burn area, age, sex, burn location, preexisting disorders), two only, deep burn area and age, have been retained as predictive factors which, when associated, allow to classify 94.47% of the patients in either survival or death group. The prognosis weight of the deep burn area (SBP) is superior to that of the total burn skin area, yet retained in most previous studies. The superficial burn area, the inhalation injuries and the preexisting disorders are not factors determining prognosis. The F equation = e(-6.0061 + (0.0829*SBP) + (0.0443*%AGE)) resulting of the logistic regression, allows a direct evaluation of the death probability. A simple linear relation can be proposed as score of severity: IG = (2 x %SBP) + age. Below 80, mortality is close to zero, above mortality increases linearly up to 210, reaching 100%. This relation must be handled cautiously when comparing the severity score of two groups of patients, just as any other severity score that uses a linear relation with the burnt area associated or not to age.