The objective of this paper is to provide a critical evaluation of the data available from existing studies concerning non-point source pollution (NPS). NPS pollution is complex and difficult to detect and manage when compared with point source pollution. To tackle its risk, it is vital to have precise simulations and estimations of NPS pollutants. Different modelling techniques applied to NPS pollution were reviewed and classified as either physically-based models or empirical. The physically-based models (White box models) can be used both for long-term and daily time steps. They require initial model data as well as watershed morphological and physiographic, which makes them complex and not easy to use. Empirical models on the other hand are called black-box models or metric models and can be used for both long-term, daily time steps with minimal data requirement and requires less skill to operate. Although their results are easy to interpret, these types of models are only suitable within the boundary of a certain domain. The findings of this review will serve as a guide to water resource planners in identifying the type of NPS model they need to apply to a particular catchment for a particular problem.