Sometimes there is a need to examine the spatial distribution of roots in studies of rice plants. For analysis of growth direction of primary roots (GDPR), a statistical method requiring an assumption of a given distribution is not available, because the distribution pattern of GDPR is not always known. A nonparametric method will be useful for analysis of such an unknown distribution. We tested the difference in distribution of GDPR with Kolmogorov-Smirnov two sample test method among three samples of rice cultivars, Nankin 11 (A), Dobashi 1 (B) and Mubo-aikoku (C). The relative cumulative frequency distribution of GDPR (Sn(x)) was calculated for each sample. When the maximum absolute difference of Sn(x) was larger than the critical value (D_α) between two samples, we concluded that the two samples have significantly different distributions of GDPR at the level of α. The result of the test indicated a difference in distribution of only A from that of B or C, though it seemed in histogram that B had a distribution pattern of GDPR intermediate beetween those of A and C. That suggests the effectiveness of nonparametric methods for precise analysis of GDPR.