Chemometric tests were carried out to better understand the multidimensional facet of starch fine structure-relationship concerning gelatinization and pasting properties. With Ward's hierarchical cluster analysis 20 long-grain rice starch samples were sorted out into three clusters based on similarities in functional properties, particularly, paste peak (PV) and final viscosity (FV). The three clusters (arbitrarily named Clusters A, B, and C) exhibited a pasting profile trend of PV,FV, PV,FV, and PV.FV, respectively. Cluster A samples were also lower in peak temperature, range and enthalpy of gelatinization, and swelling power. These attributes were associated with higher amylose content (AM), b-amylolysis limit, and percentage of B1 chains (DP13-24), but lower amylopectin weight-average molar mass (M w ) and percentage of A chains (DP6-12). A 5-variable linear discriminant function correctly predicted 85% of the Ward's cluster membership of the individual cultivars. The discriminant function included the variables A, B1, and B2 (DP25-36) chains, average chain length (ACL), and gyration radius (R z ). Fine structure variance was fully explained by a total of nine principal components, with the first three components cumulatively accounting for 74%. The leading variables included in the three rotated components pertained to amylopectin chain length distribution (A, B2, and B31 or DP!37 chains, and ACL) and amylopectin molar mass (M w , R z , and polydispersity). AM and M w were loaded most frequently in the 4-variable, best-fit linear regression models for predicting gelatinization and pasting properties. A combination of at least two fine structure variables controls the functionality of rice starch.