Acute kidney injury (AKI) is a process that can lead to renal failure. No biological markers are available for predicting the cause or prognosis of AKI. Tests that can predict which patients will need renal replacement therapy (RRT) are needed. In this chapter, we review the recent literature for proteomic analysis in AKI and identify new candidate markers to predict the need for RRT. We also used artificial neural network (ANN) analysis of urine protein data obtained by two-dimensional gel electrophoresis from 19 patients with acute tubular necrosis to identify a set of proteins that can predict whether a patient will require RRT. Ten patients were randomly selected to train an ANN algorithm. The remaining 9 patients were withheld to serve as an independent validation set. The ANN algorithm correctly predicted the renal prognosis of all 10 patients in the training set. In the validation set, the test correctly predicted the future course of renal failure in 7 of the 9 patients (78% accuracy) including 3 of 4 patients who would require RRT (75% sensitivity) and 4 of 5 who would not (80% specificity). Combinations of urine proteins can be used to predict which patients will require RRT.