Abstract: The paper aims to study the correlation and regression relationship between the harvest index and plant traits. We determined harvest index, yield per plant, plant height, 1000-grain weight, seeds pod, the main inflorescence pod length, pods per plant, effective branch position, first branch number and the main inflorescence length of 649 new combinations of Brassica uapus . The harvest index of different new combinations varied significantly. The amplitude of variation was 0.1292-0.3358, and the average was 0.2247. The proportion of harvest index which was more than 0.30 only accounted for 0.31%. The harvest index showed highly significant correlation with yield per plant, grain number, number of silique per plant, first branch number, length of main inflorescence and plant height. The results of regression analysis showed that five traits including plant height, number of silique per plant, length of main inflorescence, 1000-grain weight and grain number had greater impact on the harvest index. Thus, we can improve grain number and 1000-grain weight of breeding materials by utilizing the correlation among the traits and selecting materials that have longer silique and main inflorescence. In addition, we can select materials which are relatively higher with longer main inflorescence and more branches to improve the number of silique per plant of the bred material, to breed a combination (or variety) with higher single plant yield eventually.