Methods for predicting a response to an erythropoietic agent in a subject include providing a biological sample from the subject, and determining an amount in the sample of at least one peptide selected from the group consisting of SEQ ID NOS: 1-17. If there is a measurable difference in the amount of the at least one peptide in the sample, when compared to a control level of the same peptide, the subject is then predicted to have a good response or a poor response to the erythropoietic agent.
 This invention was made with U.S. Government support under Grant Numbers P30ES014443, DK085673-01, DK077331, and DK072085 awarded by the National Institutes of Health; Grant Number DE-FG02-05ER6406 awarded by the U.S. Department of Energy; and individual Merit Awards awarded by the Department of Veterans Affairs to Drs. Michael E. Brier and Jon B. Klein. The government has certain rights in this invention.