Fourteen cycles of mass selection for grain yield in Krug(BSK) maize (Zea mays L.) population were evaluated. Estimates of genetic components of variance were obtained for comparison with those computed from the eighth cycle of half-sib and S(,1) selection conducted in the same population. The CO and the even cycles (C2 to C14), and two sets of testcrosses (Cn x CO and Cn x B73) were evaluated in nine environments to estimate progress from mass selection. Unselected S(,1) lines derived from the original, mass, half-sib, and S(,1) selected populations wer evaluated in four environments to obtain the estimates of genetic variance;Significant, but low (0.50 q/ha), linear response for grain yield was obtained from mass selection for grain yield. Increased yield was associated with delayed silk emergence, increased stalk breakage, increased grain moisture, and greater ear height. Estimates of genetic variability for grain yield indicated a decrease of genetic variance for the S(,1) derived population and no change for mass and half-sib derived populations. For many traits, half-sib and S(,1) derived populations had similar estimates of genetic variance. The mass selected population had increased genetic variability for silk date, ear height, stalk lodging, and grain moisture. Estimates of heritability were relatively high for most traits;Although significant linear response was observed from mass selection, the mean performance of S(,1) progenies derived from the mass selected population were 6.85% lower yielding than S(,1) lines from the original population. Reduced yield for the mass selection derived lines was accompanied by delayed silking, decreased ears per plant, increased ear height, and greater stalk breakage. The S(,1) progenies from half-sib and S(,1) selected populations, however, had increased grain yield, reduced ear height, root lodging, and stalk breakage, and increased ears per plant;Predicted genetic gains computed for each trait in each population indicated that S(,1) selection was superior to mass and half-sib selection methods.