The optimal hemoglobin level in patients with hypertension or heart failure is not yet defined. The aim of the present investigation was to examine the relation of hemoglobin with cardiovascular outcomes in high-risk patients with isolated systolic hypertension (ISH) and left ventricular hypertrophy (LVH). In 1,326 patients with ISH in the Losartan Intervention For Endpoint Reduction in Hypertension (LIFE) study, hemoglobin and cardiovascular outcomes were examined using Cox proportional hazard models. Baseline hemoglobin was negatively related to rate of cardiovascular death (hazard ratio 0.81 per 1 g/dl, 95% confidence interval [CI] 0.67 to 0.98, p = 0.032) after adjusting for baseline Framingham risk score, LVH, treatment, and estimated glomerular filtration rate. Hemoglobin decreased slightly during the study and the decrease was more pronounced in the losartan group (13.9 +/- 1.3 to 13.6 +/- 1.4 g/dl) than in the atenolol group (13.9 +/- 1.2 to 13.8 +/- 1.4 g/dl). Hemoglobin as a time-varying covariate was negatively associated with rate of cardiovascular death (hazard ratio 0.75, 95% CI 0.63 to 0.90, p <0.001) and stroke (hazard ratio 0.84, 95% CI 0.72 to 0.99, p = 0.040) after adjusting for baseline Framingham risk score, LVH, treatment, and estimated glomerular filtration rate. In conclusion, in this high-risk population with ISH and LVH, lower hemoglobin at baseline was associated with higher probability of cardiovascular death, and decrease in hemoglobin over time was associated with higher probability of cardiovascular death or stroke; this effect was attenuated by treatment with losartan.