Hypophosphatemia has long been reported to be associated with sepsis and has been correlated with sepsis severity. This retrospective study was undertaken at a university hospital to determine whether severe hypophosphatemia could serve as a mortality predictor in septic patients. Charts of 6,190 septic patients who were hospitalized during one year (2001-02) were examined. Fifty-five patients were selected and were divided into 2 groups: group 1 comprised 26 patients with severe hypophosphatemia (serum inorganic phosphate (Pi) <1 mg/dl); group 2 comprised 29 patients without severe hypophosphatemia (Pi >1 mg/dl. The patients' charts were reviewed and information was collected regarding medical anamnesis, physical examination, hematological and biochemical analyses, chest x-ray, and cultures of blood and urine. The results demonstrated that 80.8% of the patients with severe hypophosphatemia died, vs 34.5% of the patients without severe hypophosphatemia (p = 0.001). Being in the severe hypophosphatemic group increased the risk of death by nearly 8-fold (odds ratio = 7.98; 95% CI = 2.3 to 27.6). These findings indicate that severe hypophosphatemia can serve as an independent mortality predictor in sepsis.