We previously reported that lower serum magnesium is associated with poorer nutrition status, impaired cellular health, and increased inflammation in peritoneal dialysis (PD) patients. The present study was designed to investigate the prognostic value of serum magnesium for mortality in PD patients. From November 2000 to July 2008, the study enrolled 62 patients, recording their demographic, clinical, and biochemical data. Patients were followed to September 2011. Mean age of the patients was 55 +/- 16 years, and in this cohort, 55% were women, 63% were African American, and 25% had diabetes. Mean serum magnesium was 1.597 +/- 0.28 mEq/L. Maximum follow-up was 10.8 years. During the follow-up period, 27 patients died (43.5%). Serum magnesium was significantly higher in the patients who survived than in those who did not (1.757 mEq/L vs. 1.515 mEq/L, p = 0.04). Patients were then stratified by enrollment magnesium. After 10.8 years of observation, cumulative survival was significantly better in patients with an enrollment serum magnesium greater than 1.6 mEq/L than in patients with an enrollment serum magnesium of 1.6 mEq/L or less (p = 0.04). Multivariate Cox regression analysis revealed that serum magnesium is a significant predictor of mortality (relative risk: 0.984; p = 0.048) after adjusting for age, race, sex, diabetes, and months on dialysis at enrollment. In conclusion, lower serum magnesium is a significant predictor of higher mortality in PD patients. Factors affecting the serum magnesium concentration in PD patients should be investigated in more detail.