Anemia and low and high levels of hemoglobin have been associated with increased mortality and morbidity. However, most studies have measured hemoglobin at only 1 time point, and few studies have considered possible reverse causation. We used data from the Women's Health Initiative, in which baseline hemoglobin was measured in 160,081 postmenopausal women and year 3 hemoglobin was measured in 75,658 participants, to examine the associations of hemoglobin concentration with total mortality, coronary heart disease mortality, and cancer mortality. Women were enrolled from 1993 to 1998 and followed for a median of 16 years. Cox proportional hazards models were used to estimate the relative mortality hazards associated with deciles of baseline hemoglobin and the mean of baseline + year 3 hemoglobin. Both low and high deciles of baseline hemoglobin were positively associated with all 3 outcomes in the total cohort. In analyses restricted to women with 2 measurements, a low mean hemoglobin level was robustly and positively associated with all 3 outcomes, after exclusion of the early years of follow-up. High mean hemoglobin was also associated with increased risk of total mortality, whereas associations with heart disease mortality and cancer mortality were weaker and inconsistent. Our results provide evidence that low and high levels of hemoglobin are associated with increased risk of mortality in otherwise healthy women.