Triage guidelines are needed to help in the decision process of intensive care unit (ICU) versus non-ICU admission for patients with diabetic ketoacidosis (DKA). Pediatric risk of mortality (PRISM) scores have long been used to assess mortality risk. This study assess the usefulness of the traditional PRISM score and adaptation of that score (PRISM-ED, which uses presentation data only) in predicting hospital stay in pediatric patients with DKA. PRISM and PRISM-ED were tested for correlation with length of stay and length of ICU stay. A medical record review was conducted for patients admitted to The Children's Hospital of Alabama with DKA during an 18-month period (n = 79). Two scores were calculated for each study entrant: PRISM using the worst recorded values over the first 24 hours and PRISM-ED using arrival values. Median scores, median test, and Spearman rank correlations were determined for both tests. Median PRISM scores were PRISM = 11 and PRISM-ED = 12; Median PRISM and PRISM-ED scores for patients admitted to the ICU were less than median scores among floor-admitted patients: [table: see text] Spearman rank correlations were significant for both scores versus total stay: PRISM, rs = 0.29; P = 0.009; PRISM-ED, rs = 0.60, P < 0.001. Also, correlations were significant for both scores versus ICU stay: PRISM rs = 0.22, P = 0.05; PRISM-ED, rs = 0.41, P < 0.001. Triage guidelines for ICU versus floor admission for DKA patients could have significant economic impact (mean ICU charge = $11,417; mean charge for floor admission = $4,447). PRISM scores may be an important variable to include in a multiple regression model used to predict the need for ICU monitoring.