Twenty seven cycles of divergent mass selection for ear length were completed in the Iowa Long Ear Synthetic (BSLE) cultivar. The original population, each third cycle of selection for increased and decreased ear length, their respective S1 lines, and their respective crosses were used to estimate direct and correlated responses to selection and the effect of selection on inbreeding depression and heterosis. This experiment was conducted in 1995 and 1996 at five locations of Iowa. 300 unselected S1 lines were obtained from the original population and from the cycles of selection of both short- and long-ear subpopulations to determine the effect of selection on the genetic variability for ear length and correlated traits. This experiment was conducted in 1995 and 1996 at two locations of Iowa. Divergent mass selection was highly effective to modify ear length. After twenty seven cycles of selection, ear length was increased 1.4% (b1 = 0.27 ± 0.03) per cycle of selection for longer ears and decreased 1.9% (b1 = -0.37 ± 0.03) per cycle of selection for shorter ears. Selection for shorter ear was accompanied by a significant decrease of grain yield of 1.7% (b1 = -0.08 ± 0.01) per cycle of selection, whereas selection for longer ear did not significantly modify grain yield. Selection for shorter ear was also accompanied by a significant increase of kernel-row number (b1=0.11±0.01) and ear diameter (b1=0.10±0.01) whereas selection for longer ear significantly reduced the expression of kernel row number )b1=0.06±0.01) and ear diameter (b1=-.20±0.01). Hence, improvement in grain yield in the longer-ear subpopulation was canceled by a significant reduction of the kernel-row number and ear diameter. The rate of inbreeding depression of ear length and yield remained constant during the course of selection in the longer-ear subpopulation, but it was gradually reduced with continued selection in the shorter-ear subpopulation. No heterosis was observed in the crosses between corresponding cycles of selection. Estimates of genetic variability among S1 progenies showed no evidence that genetic variance is been exhausted. Hence, further progress in selection is expected in both short- and long-ear subpopulations.