Abstract: In this study, 12 maize populations had been improved using improved S1 selection approach, aiming to increase the efficiency of the improvement of maize populations. The results showed that, number of rows per spike, number of grains per row and 100-seed weight were the three component traits of maize yield. The highest genetic increment was founded in Mengqun2, and followed by Mengqun4 and mengqun1 etc, 7 maize populations. Negative genetic increment was founded in Mengqun3 and 3 foreign maize populations. Some changes had been found in spike, plant traits and genetic diversity of maize populatio. Based on our results, we concluded that the first-class plants could be selected from Mengqun2, Mengqun4, Zhongzong7, Meng C qun and Meng A qun directly to filter out lines by selfing because of the higher yield genetic increment and significant improvements of other agricultural genetic traits of these maize populations. Comparing with those five populations, the improvement potential of other maize populations was limited because of the lower genetic increment.