Specialized treatment planning software systems are generally required for neutron capture therapy (NCT) research and clinical applications. The standard simplifying approximations that work well for treatment planning computations in the case of many other modalities are usually not appropriate for application to neutron transport. One generally must obtain an explicit three-dimensional numerical solution of the governing transport equation, with energy-dependent neutron scattering completely taken into account. Treatment planning systems that have been successfully introduced for NCT applications over the past 15 years rely on the Monte Carlo stochastic simulation method for the necessary computations, primarily because of the geometric complexity of human anatomy. However, historically, there has also been interest in the application of deterministic methods, and there have been some practical developments in this area. Most recently, interest has turned toward the creation of treatment planning software that is not limited to any specific therapy modality, with NCT as only one of several applications. A key issue with NCT treatment planning has to do with boron quantification, and whether improved information concerning the spatial biodistribution of boron can be effectively used to improve the treatment planning process. Validation and benchmarking of computations for NCT are also of current developmental interest. Various institutions have their own procedures, but standard validation models are not yet in wide use.