AbstractSimple and multiple linear regressions were applied to the development of models for predicting biodegradability of chemicals from molecular connectivity indexes. Indexes were calculated using the program CFUNC, on the DECsystem 10 computer at the National Institutes of Health, Bethesda, Maryland. Biodegradation data from test systems containing natural media or inocula were obtained from the literature for several homologous series of organic compounds; within each series, all compounds had been tested under the same conditions. The following one‐variable models were among those obtained: for 2,4‐D alkyl esters, log rate constant versus 2Xv, r2 = 0.954 (n = 6); for N‐3‐chlorophenylcarbamates, log percent degraded versus 4Xp/c, r2 = 0.971 (n = 7); for dialkyl ethers, log percent theoretical oxygen demand (%ThOD) versus 2Xv, r2 = 0.974 (n = 6); for dialkyl phthalate esters, rate constant versus 2X, r2 = 0.938 (n = 12); for aliphatic acids, %ThOD versus 4Xc, r2 = 0.895 (n = 10). Two‐variable models substantially improved results for aliphatic alcohols and acids. These results indicate that the apparent effect of alkyl chain branching on biodegradability can be quantitatively described by connectivity indexes, especially 2X, 3Xc, 4Xc and 4Xp/c, 3Xp also appeared frequently, especially in two‐variable models, suggesting a possible role for molecular shape as a determinant of biodegradability. The results suggest that molecular topology holds promise for future research in quantitative structure‐biodegradability relationships.