以《中国稻种资源目录》中的 16 791份中国粳稻地方种为材料 ,根据完全随机、变种类型 (丁颖分类法 )下随机和系统聚类 3种取样方法和 4种核心样品规模 (12 .5 %、10 %、7.5 %和 5 % )设计 12个核心样品 ,以 8个编目性状的平均数变幅、方差和变异系数加以评价。结果表明 :(1)变种类型下系统聚类取样方法是构建中国粳稻地方种资源核心样品的适宜方法 ;(2 )样品规模 5 %的核心样品可以有效地代表中国粳稻地方种资源的总体样品 Based on 16 791 Chinese japonica rice landraces in the “China Rice Germplasm Resources Catalog”, according to three sampling methods of random and systematic clustering under completely random and variable types (Ding Ying classification) and four core sample sizes .5%, 10%, 7.5%, and 5%), 12 core samples were designed and evaluated using the mean number of 8 cataloged traits, variance and coefficient of variation. The results showed that: (1) The systematic clustering sampling method under varietal type was the suitable method for constructing the core sample of japonica rice endemic species in China; (2) The core sample with 5% sample size could effectively represent the overall sample of local japonica rice species