Simultaneous blood and saliva samples were collected for determination of the relationship between serum and saliva erythropoietin (EPO) concentrations in 12 adults (Group I), 15 full-term neonates (Group II), and 11 premature infants (Group III). Saliva was collected with a modified sputum-collecting tube combined with a vacuum suction pump. Serum and saliva EPO concentration was measured by enzyme-linked immunosorbent assay. The relationship between serum and saliva concentrations was explored using (1) regression analysis and (2) serum-to-saliva ratio. Salivary concentrations approximated 15 to 30% of the serum concentrations based on the serum-to-saliva ratios. Significant correlation was observed between serum and salivary concentrations in each group (p < 0.05). The regression analyses produced formulas for predicting serum EPO concentrations from saliva EPO concentrations which seemed to fit the data well. The resulting formula is surprisingly consistent with that derived from Group I. The ratio models seem to fit the data well as regression models. From the results it is concluded that the use of salivary samplings for serum EPO in adults, full term and premature infants may be a possible alternative method to blood samplings.