The CND0/2 molecular orbital method is used to predict and explain the barriers to internal rotation in a number of molecules. The observed 3: 2: 1 ratio of the barriers in ethane, methylamine, and methyl alcohol is approximately reproduced as are most trends in the barriers of fluoro-substituted propenes; however, the calculated trends for fluoro-substituted ethanes are incorrect. The barriers in H202, F202, N2H4, N2F4, and NH20H and the effect of geometry optimization on these barriers are also discussed. The major source of the barriers in the first group of molecules is predicted to be due primarily to nonbonded interactions across the axial bonds, while interactions between the lone pairs on the axial bonds are found to be important in the latter group. It is concluded that CNDO will be applicable to further barrier studies only if nonbonded interactions involving highly electronegative atoms (e.g., F) are unimportant.