A positive association between root morphology and herbage yield in alfalfa (Medicago sativa L.) has been reported previously. To further investigate this association, we created populations that differ in root morphology within four unrelated experimental germplasm sources. Two germplasm sources were divergently selected for lateral root number, and two sources underwent divergent selection for fibrous root mass followed by divergent selection for lateral root number. Selected and unselected populations from all germplasm sources were evaluated for herbage yield, root morphology, fall dormancy response, and disease resistance. Herbage yield was evaluated using eight replicates of a randomized complete block design with a split-plot arrangement of fertilizer rates (0 and 200 kg N ha-1) as whole plots and alfalfa populations as subplots. Experiments were established twice at each of two locations in May 1994. Two herbage yield harvests were recorded from one experiment at each location, and plots were dug and evaluated for root traits in fall 1994. Herbage yields were taken from the other experiment at each location twice in 1994 and four times in 1995, and again plots were dug and evaluated for root traits in fall 1995. All populations were evaluated for fall dormancy response in 1994 and disease resistance in 1995 according to standard protocols. Populations selected for more fibrous or lateral roots had greater herbage yield than populations selected for no or few fibrous or lateral roots in all four germplasm sources. No differences in root size or weight, dormancy, or disease resistance were found between fibrous or branch-rooted vs. taprooted populations. Selection for fibrous and lateral roots within these alfalfa germplasms increased herbage yield in the tested environments.