Background: The prognostic importance of hemoglobin is controversial. We investigated the prognostic importance of baseline and in-treatment hemoglobin in the LIFE study. Methods: Eight thousand one hundred ninety-four LIFE patients with hypertension and left ventricular hypertrophy with available baseline hemoglobin measurements were randomized to losartan- or atenolol-based treatment and followed for 4.8 years for end points of all-cause mortality and composite of cardiovascular death, nonfatal stroke, or nonfatal myocardial infarction. Results: U-shaped relations were observed between deciles of baseline hemoglobin and all-cause mortality and the composite end point. In univariate Cox models, baseline hemoglobin in the lowest gender-specific decile (women/men: <12.5/13.4 g/dL) was associated with all-cause mortality (hazard ratio [HR] 2.01, 95% CI 1.64-2.64) and the composite end point (HR 1.53, 95% CI 1.27-1.85, both P < .001), whereas hemoglobin in the highest gender-specific decile (women/men: ≥15.0/16.2 g/dL) was not. The decrease in hemoglobin was higher (P < .001) in patients allocated to losartan- (14.3-13.8 g/dL) versus atenolol-based treatment (14.3-14.0 g/dL). In Cox models with the same gender-specific definitions for high and low hemoglobin as time-varying covariates with adjustment for treatment allocation and established risk factors and diseases, hemoglobin in the lowest decile was associated with higher rates of all-cause mortality (HR 3.03, 95% CI 1.89-4.85, P < .001) and the composite end point (HR 1.36, 95% CI 1.08-1.71, P < .01), whereas hemoglobin in the highest decile was not. Conclusions: After adjusting for other risk factors, relatively low, but not high, hemoglobin during antihypertensive treatment was associated with higher incidence of all-cause mortality and the composite end point.