To evaluate varieties in regional trials, breeders would analyze the synthetical performances of the varieties. The methods of ANOVA and MANOVA are not preferable because of the frequent existence of missing data.In this study, we utilized the MINQUE(1) approaches proposed by ZHU et al and analyzed 6 agronomic traits of 15 hybrid rices in 1991 and 1992. These traits and their coefficients were: yield (0.40), period duration (PD, 0.25), rice quality (RQ, 0.20), resistance to bacterial blight (RBB, 0.075), resistance to neck blast (RNB, 0.0375) and resistance to rice planthopper (RRP, 0.0375). The estimates of variance indicated that the year×location elect was the main factor to, yield followed by the genotype×location effect. The location effects were controlled mainly PD, RNB and RBB. Rice quality was affected largely by residual effects and secondarily by genotype×year and year×location effects. RRP was controlled mostly by genotype×year and year effects. Covariance compoments between yield and o To evaluate varieties in regional trials, breeders would analyze the synthetical performances of the varieties. The methods of ANOVA and MANOVA are not preferred because of the frequent existence of missing data. In this study, we utilized the MINQUE (1) solutions proposed by ZHU et al and analyzed 6 agronomic traits of 15 hybrid rices in 1991 and 1992. These traits and their coefficients were: yield (0.40), period duration (PD, 0.25), rice quality (RQ, 0.20), resistance to bacterial blight , 0.075), resistance to neck blast (RNB, 0.0375) and resistance to rice planthopper (RRP, 0.0375). The estimates of variance indicated that the year × location elect was the main factor to, yield followed by the genotype × location effect. The location effects were controlled mainly PD, RNB and RBB. Rice quality was affected largely by residual effects and secondarily by genotype × year and year × location effects. RRP was controlled mostly by genotype × year and year effects. Covariance compoments b etween yield and o