No / We investigated five different methods for evaluating the electrostatic interaction between atoms in force field calculations on organic solids. Atomic charges and multipoles were obtained by fitting them to the molecular electrostatic potential, calculated in vacuum with an ab initio quantum mechanical method. Multipole moments were derived using three schemes, differing in the order in which the monopoles, dipoles and quadrupoles were fitted. For comparison, Gasteiger charges were also calculated. Using these electrostatic models, the lattice parameters and the molecular geometry of 48 organic crystals were optimised with the DREIDING force field. During the optimisation, the atomic multipoles were rotated with their local environment to account for molecular flexibility. For comparative reasons, rigid-body optimisations were performed on a subset of structures. The results were analysed in terms of structural parameters of the lattice and the molecules, and, for the ten polymorphic systems present in the test set, in terms of relative stability. On average, the multipole methods were not superior to the point charge methods for the full optimisation. For rigid molecules, however, the multipole models gave a substantial improvement in lattice parameters. No evidence was found that parameters for van der Waals energies need to be refitted for a specific electrostatic model. Energy differences between polymorphs were less than 5 kcal mol¿1 in eight out of ten cases, independent of the electrostatic model used. The results show that our use of distributed multipoles to describe the intra-molecular as well as inter-molecular electrostatic interactions does lead to an improvement in accuracy for rigid molecules, but not for flexible molecules. The investigation shows that accurate descriptions of the intra-molecular as well as the inter-molecular energies are crucial for the successful optimisation of crystal structures of organic solids.