Background: Despite the frequent use of continuous renal replacement therapy (CRRT) in the management of acute renal failure (ARF) in the critically ill, predictors of mortality remain unclear. Methods: A registry of all patients initiated on CRRT at a single institution was assembled over an 18-month period, and a subsequent cross-sectional analysis of selected variables was conducted for associations with mortality. Predictors evaluated were age, gender, diagnosis of sepsis, Apache II score, days between ARF diagnosis and initiation of CRRT, creatinine at initiation of CRRT, change in creatinine from baseline and admission to initiation of CRRT, setting of ARF, and prescribed CRRT dose. The principal outcome was mortality at 30 days. Results: Eighty-one individuals met inclusion criteria. Overall mortality for the study was 50.2%. The mean elevation in creatinine from admission to initiation of CRRT was 1.6 mg/dL (141.4 jumol/L) in those who lived and 2.6 mg/dL (229.8 μmol/L) in those who died (P = 0.023). Patients admitted with normal renal function who developed ARF while in the hospital had mortality of 56.3%. When available, patients with abnormal renal function at presentation were further classified by either abnormal or normal preadmission creatinine. These patients had mortality of 31.3% and 83.3%, respectively. These differences in mortality were statistically significant. Conclusions: Increased mortality was significantly associated with the magnitude of change in serum creatinine between admission and initiation of CRRT. Also, patient ARF classification was significantly associated with mortality.