The molecular geometries of biphenyl in minimum-energy twisted, copolanar, and perpendicular conformations have been optimized, with D2, D2h, and D2d symmetry restrictions imposed, by ab initio molecular orbital calculations using polarized basis sets. In the geometry optimization, phenyl rings are kept planar. The twist angles of the conformer of minimum energy calculated at HF/6-31G* and HF/6-31G** levels are 46.13 and 46.26-degrees, respectively, which are very close to the recently reported value of 45.41-degrees obtained by an HF/6-31-G level calculation and the value of 44.4 +/- 1.2-degrees from electron diffraction experiments. The calculated bond distances and angles are also close to those from HF/6-31G calculations and those from experiments. The internal rotational barrier heights at 0-degrees (DELTA-E0) and at 90-degrees (DELTA-E90) calculated at the HF/6-31G* level are 3.28 and 1.48 kcal/mol, respectively. Those at the HF/6-31G** level are 3.33 and 1.51 kcal/mol, respectively. They are very close to DELTA-E0 = 3.17 and DELTA-E90 = 1.62 kcal/mol from recent HF/6-31G level calculations. DELTA-E0 and DELTA-E90 calculated at the MP4(SDQ)/6-31G*//HF/6-31G* level are 3.47 and 1.58 kcal/mol, respectively. While the DELTA-E90 value from these calculations is close to DELTA-E90 = 1.6 +/- 0.5 kcal/mol from electron diffraction experiments, the DELTA-E0 value from these calculations is much higher than DELTA-E0 = 1.4 +/- 0.5 kcal/mol from the same experiments. The internal rotational potential calculated at the MP4(SDQ)/6-31g*//HF/6-31G* level is shallow in the region near the minimum but has a steep slope in the region of phi = 0-30-degrees. The equation V(phi) = 1/2V2(1-cos 2-phi) + 1/2V4(1-cos 4-phi) + C, which was used for the estimation of the internal rotational potential from the experimental measurement, has been fitted to the calculated potential in the region near the minimum. DELTA-E0 estimated from the fitted equation is much lower than the value from the ab initio calculation, which shows that this equation is not appropriate to estimate DELTA-E0 from the shape of the potential near the minimum, if the shape of the potential is similar to that obtained by the ab initio method.