BACKGROUND: The relationship between low hematocrit and contrast-induced nephropathy has not been investigated. METHODS: Of 6,773 consecutive patients treated with percutaneous coronary intervention, contrast-induced nephropathy (an increase of >/=25% or >/=0.5 mg/dL in preprocedure serum creatinine, at 48 hours postprocedure) occurred in 942 (13.9%) patients. RESULTS: Rates of contrast-induced nephropathy steadily increased as baseline hematocrit quintile decreased (from 10.3% in the highest quintile to 23.3% in the lowest quintile) (chi(2) for trend, P < 0.0001). Stratification by baseline estimated glomerular filtration rate (eGFR) and baseline hematocrit showed that the rates of contrast-induced nephropathy were the highest (28.8%) in patients who had the lowest level for both baseline eGFR and hematocrit. Patients with the lowest eGFR but relatively high baseline hematocrit values had remarkably lower rates of contrast-induced nephropathy (15.8%, 12.3%, 17.1%, and 15.4% in 2nd, 3rd, 4th, and 5th quintiles of baseline hematocrit, respectively) (P < 0.0001). The rates of contrast-induced nephropathy increased with increment in change in hematocrit. Patients in the lowest quintile of baseline hematocrit with absolute hematocrit drop >5.9% had almost doubled rates of contrast-induced nephropathy compared with patients with hematocrit change <3.4% (38.1% vs. 18.8%, respectively) (P < 0.0001). By multivariate analysis, lower baseline hematocrit was an independent predictor of contrast-induced nephropathy; each 3% decrease in baseline hematocrit resulted in a significant increase in the odds of contrast-induced nephropathy in patients with and without chronic kidney disease (11% and 23%, respectively). When introduced into the multivariate model instead of baseline hematocrit, change in hematocrit also showed a significant association with contrast-induced nephropathy. CONCLUSION: Lower hematocrit is an important risk factor for contrast-induced nephropathy. Whether correcting the hematocrit prepercutaneous coronary intervention might decrease the rates of contrast-induced nephropathy should be addressed in a prospectively designed trial.