The reported data of biodegradability of variously substituted 25 sulfonated azo dyes were divided into two classes: 8 dyes designated “rapidly biodegradable” and 17 dyes “not rapidly biodegradable”, by applying principal components analysis (PCA). The biodegradability of the dyes was modeled with the discriminant technique of two-value regression analysis based on structural descriptors. A statistically significant and biologically meaningful regression model that gives perfect classification was obtained, and the results were compared with the previous qualitative interpretation of the substituent effects on biodegradation. The coefficients of the regression model revealed a significant interaction between ring substituents for biodegradability. The results facilitate understanding of the mechanism of degradation.