Illicit autologous blood transfusion to improve performance in elite sport is currently undetectable, but the stability of longitudinal profiles of an athlete's hemoglobin mass (Hbmass) might be used to detect such practices. Purpose: Our aim was to quantify within-subject variation of Hbmass in elite athletes, and the effects of potentially confounding factors such as reduced training or altitude exposure. Methods: A total of 130 athletes (43 females and 87 males) were measured for Hbmass an average of six times during a period of approximately 1 yr using carbon monoxide rebreathing. Linear mixed models were used to quantify within-subject variation of Hbmass and its associated analytical and biological components for males and females, as well as the effects of reduced training and moderate altitude exposure in certain athletes. Results: The maximum within-subject coefficient of variation (CV) for Hbmass was 3.4% for males and 4.0% for females. The analytical CV was ∼2.0% for both males and females, and the long-term biological CV, after allowing for analytical variation, was 2.8% for males and 3.5% for females. On average, self-reported reduced training resulted in a 2.8% decrease in Hbmass and altitude exposure increased Hbmass by 1.5% to 2.9%, depending on the duration and type of exposure. Conclusions: The within-subject CV for Hbmass of ∼4% indicates that athletes may experience changes up to ∼20% with a 1-in-1000 probability. Changes of this magnitude for measures taken a few months apart suggest that Hbmass has a limited capacity to detect autologous blood doping. However, changes in Hbmass may be a useful indicator when combined with other measures of blood manipulation.