This paper presents data on hemoglobin concentration in a rural Andean sample at 3,800-3,900 m and incorporates them into a review intended to evaluate possible sources of the range of variation in mean hemoglobin concentration among samples obtained at high altitude. Between 3,400 and 4,000 m, rural Himalayan highlanders average 1.4 gm/dl lower mean hemoglobin concentration than rural Andean highlanders. With respect to potential causes of anemia, it is concluded that the relatively low values of rural Himalyan populations are not explicable by lower hypoxic stress or different techniques of obtaining and analyzing blood samples and are probably not explicable by nutritional deficiency and disease. With respect to potential causes of polycythemia within Andean populations, it is concluded that the somewhat higher values of some mining and urban samples of Andean higlanders may not be due to the mining occupation per se but may be due partly to the inclusion of European and mestizo (with at most 500 years of high-altitude ancestry) along with Amerindian highlanders (with millenia of high-altitude ancestry) as well as to the inclusion of highlanders living well above their own habitual altitudes of residence. The Andean polycythemia is probably not due to obesity, high androgen levels, or frequent intermittent hypoxemia during sleep. The effect of heavy smoking cannot be evaluated. Further work on hematological adaptation to high altitude must pay special attention to sample characteristics.