Although plasma neutrophil gelatinase-associated lipocalin (NGAL) is a promising biomarker for early detection of acute kidney injury, its ability to predict recovery is unknown. Using RIFLE criteria to define kidney injury, we tested whether higher plasma NGAL concentrations on the first day of RIFLE-F would predict failure to recover in a post hoc analysis of a multicenter, prospective, cohort study of patients with community-acquired pneumonia. Recovery was defined as alive and not requiring renal replacement therapy during hospitalization or having a persistent RIFLE-F classification at hospital discharge. Median plasma NGAL concentrations were significantly lower among the 93 of 181 patients who recovered. Plasma NGAL alone predicted failure to recover with an area under the receiver operating characteristic curve of 0.74. A clinical model using age, serum creatinine, pneumonia severity, and nonrenal organ failure predicted failure to recover with area under the curve of 0.78. Combining this clinical model with plasma NGAL concentrations did not improve prediction. The reclassification of risk of renal recovery, however, significantly improved by 17% when plasma NGAL was combined with the clinical model. Thus, in this cohort of patients with pneumonia-induced severe acute kidney injury, plasma NGAL appears to be a useful biomarker for predicting renal recovery.