SETTING: Special Care Ward (SCW) of the Dhaka Hospital of icddr,b, Bangladesh. OBJECTIVE: To evaluate the clinical and laboratory predictors of death in under-five children with clinically defined sepsis presenting with diarrhea. METHODS: We prospectively enrolled all the diarrheal children (n = 151) aged 0 to 59 months with clinical sepsis admitted in the SCW during September’2007 through December’2007. Comparison was made between deaths (n = 23) and survivors (n = 128). Sepsis is defined as presence of inflammation [abnormal WBC count (>11 × 109/L or, <4 × 109/L or, band and neutrophil ratio ≤0.10] plus presence or presumed presence of infection with thermo-instability [hypo (≤35.0℃) or hyperthermia (≥38.5℃)], tachycardia, tachypnea, and/or the indications of altered organ function (altered mental status and bounding pulse) in the absence of clinical dehydration or after correction of dehydration. RESULTS: The median (inter-quartile range) age (months) of the children who survived and died was comparable [4.0 (2.0, 12.0) vs. 1.5 (0.8, 10.0); p = 0.703]. In the logistic regression analysis, after adjusting for potential confounders, such as abnormal WBC count, use of intravenous fluid, patient with fatal outcome more often presented with hypernatremia (odds ratio = 16.48, 95% confidence interval = 2.21 - 123.12; p = 0.006), lobar consolidation (odds ratio = 19.9, 95% confidence interval = 2.99 - 132.80; p = 0.002), hypoxemia (odds ratio = 14.78, 95% confidence interval = 1.38 157.90; p = 0.026) and severe under-nutrition (odds ratio = 7.57, 95% confidence interval = 1.24 - 46.11; p = 0.028). CONCLUSIONS: Our data suggest that children under-five with clinical sepsis who present with lobar pneumonia, hypoxaemia, severe acute malnutrition and hyperntaemia are at higher risk of death and identification of these simple factors may help clinicians to take prompt initiative for the aggressive management of such children especially in a resource-limited setting like Bangladesh.