Having access to highly accurate potential energy surfaces (PESs) is a prerequisite for the ab initio calculation of accurate spectroscopic constants and molecular properties. The construction of such surfaces, however, is a formidable task. It not only requires solving of the time‐independent, nonrelativistic, clamped‐nuclei, electronic Schrödinger equation of the molecular
n
‐electron system at the highest possible level of approximation, but it also requires inclusion of small energy terms that are often neglected at other occasions. Such terms are, for example, due to nonadiabatic and relativistic effects (scalar and nonscalar). Nevertheless, the largest sources of error in calculating PESs are most often due to the inadequate description of the nonrelativistic electronic wave function of the molecular
n
‐electron system in the Born–Oppenheimer approximation. This article is concerned with these sources of error, which are related to the truncation of the one‐electron basis set of atomic functions and to the incomplete expansion of the wave function in a basis of
n
‐electron functions. In particular, these errors diminish only extremely slowly with the extension of the one‐electron basis set when the wave function is expanded in terms of
n
‐electron functions that are (antisymmetrized) products of one‐electron functions (orbitals). Coupled‐cluster theory in conjunction with explicitly correlated two‐electron geminals (instead of the more commonly used products of two orbitals) appears to be a promising approach toward obtaining an adequate description of the clamped‐nuclei, nonrelativistic electronic wave function and, thus, for the calculation of highly accurate PESs, from which accurate spectroscopic constants and molecular properties may be obtained.