AbstractFive experiments were conducted to determine the effects of herbicides on seedling injury and on weed control in the establishment of alfalfa (Medicago sativa L.), ladino clover (Trifolium repens L.), and crownvetch (Coronilla varia L.) in North Carolina. Six treatments with commercial herbicides were included in the series of tests, and a number of other commercial and experimental herbicides were included in one or more of the tests. Two of the five tests were spring seeded to determine the feasibility of using herbicides for successful spring seeding in the area as opposed to the standard practice of fall seeding. Each herbicide was tested in either a preplant incorporated, preemergence, or postemergence application. Herbicide effects were measured in terms of legume stands, vigor, and injury ratings. All the herbicides caused some injury but certain herbicides were only slightly injurious while giving excellent weed control. In prepllant incorporated applications, S‐propyl dipropylthiocarbamate (vernolate) and S‐ethyl dipropylthiocarbamate (EPTC) provided excellent weed control and insignificant injury to all legumes. The higher of two rates of vernolate and EPTC were only slightly more injurious to the legumes than the lower rates. Preplant incorporated treatments of N‐butyl‐N‐ethyl‐a,a,a‐triflouro‐2,6‐dinitro‐p‐toluidine (benefin), 4‐(methylsulfonyl)‐2,6‐dinitro‐N,N‐ dipropylaniline (nitralin), and a,a,a‐triflouro‐2,6‐dinitro‐N,N‐dipropyl‐p‐toluidine (trifluralin) generally gave low injury, but effects were variable for tests and legume species, especially at the higher rate. Postemergence treatment with 2‐sec‐butyl‐4,6‐dinitrophenol (dinoseb) caused little injury in certain tests but resulted in severe defoliation in others, with the least injury to alfalfa and the most to crownvetch. None of the herbicides used in preemergence applications gave good weed control without severe legume injury. There were no readily apparent differences between fall and spring effects of the herbicides. Injury to legumes in one test on sandy soil was generally greater than in the four tests on clay soil.