We have used 75% to 90% pure murine erythroid colony-forming units (CFU- E) to delineate the processes and factors underlying their maturation. These CFU-E form 32 cell colonies and are drawn from what we term generation I of a six-generation long maturation sequence (Landschulz et al, Blood 79:2749, 1992). Applying assays of 59Fe-heme biosynthesis and colony numbers as measures of maturation and analyses of DNA degradation as an index of programmed cell death, we find that (1) erythropoietin (Epo) enhances maturation throughout most of its course; (2) Epo first seems able to forestall DNA degradation when CFU-E reach generation II; (3) the processes that Epo elicits thereafter start to persist when Epo is withdrawn; (4) insulin-like growth factor I (IGF-I) also forestalls DNA breakdown, but later loses effectiveness; (5) IGF-I adds little to maturation when Epo levels are high, but when Epo levels are low, enhances it substantially; and (6) for maturation to be entirely optimal, an unidentified serum factor(s) is probably required when Epo levels are high and is certainly needed when Epo levels are like those in normal animals. Quantitatively, about 40% of optimal in vitro erythropoiesis at normal Epo levels depends on Epo alone, another 30% or less on the addition of IGF-I, and the remaining 30% or more on the addition of unidentified serum factor(s). Applied together, these three or more factors lead to two-thirds of the maximum maturation realized with saturating Epo levels. Because we also find that heme accumulated in CFU-E culture can closely approach levels in red blood cells, we suppose that our conclusions apply as well to CFU-E maturation in vivo.