We evaluated nine systems of weed control over a 3-year period for cotton (Gossypium hirsutum L.) tolerance and control of weeds on Tifton loamy sand soil. Cultivation with sweeps did not control weeds effectively, which allowed weed competition to reduce cotton yields. Five cultivations plus an annual average of 185 hr/ha hand-hoeing was required to keep cotton relatively free of weeds for 7 to 8 weeks after planting. The use of herbicides offered certain economic advantages over systems using only cultural practices in controlling weeds. A preplant application of trifluralin (α,α,α-trifluoro-2,6-dinitro-N,N-dipropyl-p-toluidine) reduced annual hand-hoeing requirements by 66 hr/ha. The level of weed control at harvest time increased each year in all systems that used a herbicide program, but not where only cultivation or hand-hoeing was used. The continuous use of specific weed-control systems shifted weed populations. Systems that achieved 100% control by the end of the third year included all of the following: (a) a preplanting or at planting treatment of trifluralin or EPTC (S-ethyl dipropylthiocarbamate) plus (b) a preemergence application of fluometuron [1,1-dimethyl-3-(α,α,α-trifluoro-m-tolyl)urea], (c) two directed postemergence sprays with MSMA (monosodium methanearsonate), and (d) diuron [3-(3,4-dichlorophenyl)-1,1-dimethylurea]. Large crabgrass [Digitaria sanguinalis (L.) Scop.] and Florida pusley (Richardia scabra L.) were nearly eliminated where we used trifluralin. Systems that did not include a postemergence application of MSMA allowed common cocklebur (Xanthium pennsylvanicum Wallr.), Florida beggarweed [Desmodium tortuosum (Sw.) DC], and other weeds to emerge and remain in the crop row, reducing yield of lint cotton. Cotton yield appeared to be directly related to the level of early-season weed control. We did not detect biologically active soil residues at harvest time each year. Our weed-control systems did not affect cotton fiber length, strength, or color.