Abstract: Abstract: The differences in water and nutrients uptake by rice induced by tillage regime were studied to better understand the mechanism for rice growth. This study was carried out at the long-term purple soil fertility monitoring station established by the Ministry of Agriculture in 1990. Three tillage regimes including Flooded Paddy Field (FPF), Ridge with No-Tillage (RNT), and Conventional Tillage (CT) were selected to study tillage effects on nitrogen, phosphorus and potassium concentrations and carbon isotope composition in rice plants. The rice yield and samples from roots, stalks, leaves, tassels were collected for 2 rice-growing seasons at rice mature stage, from 2016 to 2017. The Principal Component Analysis (PCA) and Multiple Linear Regression (PCA-MLR) method package for rice yield was originally developed under different tillage modes to simulate the relationship of rice yield with water and nutrient elements, including carbon isotope composition, root characteristic, nitrogen, phosphorus, and potassium concentrations. Results showed that rice yields for RNT were significantly higher than those for FPF and CT (P<0.05). The carbon isotope composition of rice leaves for RNT was significantly lower than for FPF and CT for 2 years observed. Pearson's correlation analysis was used to analyze the relationship between rice yield and selected parameters. Significant negative relationships were observed between rice yield and carbon isotope composition of roots, stalks, leaves, tassels at rice mature stage, which were ?0.702, ?0.734, ?0.572, and ?0.711. The nitrogen, phosphorus, and potassium concentrations of rice tassels were positively correlated with rice yield, which was 0.411, 0.432, and 0.529, respectively. And also, significant positive relationships were observed between rice yield and root characteristics, which were 0.538, 0.624, 0.450, and 0.710 to total root length, root surface area, total root volume, and average diameter of root. The Principal Component Analysis (PCA) was conducted on the analysis of selected parameters. As a result, the Principal Components (PCs) could explain the cluster of correlated variables in groups. PCA results showed that the 6 principal components extracted from 20 selected parameters, contributed about 80% to the total variability. The principal component analysis indicated that the most important variables in explaining the 30.1% of total variations retained by PC1, which included the quality parameters carbon isotope composition, P, K in tassels and K in leaves, mainly by carbon isotope composition; PC2 explained the 12.18% of total variability, which included the quality parameters composed of N in stalks and leaves; PC3 was composed of roots characteristics, which was 11.78%; and PC4 was composed of P in stalks and leaves, which was 10.05%. The work used PCA-MLR to analyze the linear relationship of rice yield to selected various elements. The retained principal components extracted by PCA were used in regression analysis with rice yield. The determination coefficients R2 of models were 0.67, 0.73, and 0.97 related to FPF, RNT, and CT implied a fair accuracy of PCA-MLR models. PCA-MLR analysis implied rice yield was mainly impacted by selected PC1 and PC4 for RNT, where the regression coefficient was ?0.662 and 0.437. Significant relationships were observed between rice yields, carbon isotope composition, and phosphorus contents in rice plants for RNT and FPF by PCA-MLR analysis. Whereas for CT, rice yield was better related to roots characteristics and nitrogen contents in rice plants. Carbon isotope composition in rice plant might be a better parameter to predict rice yields for FPF and RNT, rather than for CT. Therefore, RNT could be a good technology to improve soil fertility and rice yield in the future with the support of mechanized planting.