Alfalfa (Medicago sativa L.) germplasm pools developed by two cycles of phenotypic recurrent selection for forage dry matter production and forage reduced N concentration were evaluated under glasshouse conditions. Selections were based on previously described protocols that involved independent culling levels for dry matter and reduced‐N concentration in forage produced by each genotype grown sequentially on N2 and NH4NO2 as N sources. A single Rhizobium genotype (R. meliloti Dang, strain 102F28) was used in all selection and test procedures, and plants were evaluated at a single morphological stage during the two cycles of selection. In two nondormant germplasm sources, ‘Hairy Peruvian’ and ‘African,’ significant (P ≤ 0.01) increases were made in both forage dry matter production and reduced‐N concentration. Two cycles of selection for high dry matter production and high reduced N concentration increased dry matter 52% and reduced N concentration 5.2% averaged over both N sources in Hairy Peruvian. Results from African showed 17 and 5.2% increases, respectively, in the same parameters. Selection for average dry matter production and either high or low reduced N concentration resulted in significant (P ≤ 0.01) changes in reduced N concentration. Only selection for high dry matter production and high reduced N concentration, however, increased total forage N.