Tropical communities in the developing world depend heavily on riverine systems for their socioeconomic development. However, these resources are poorly protected from diffuse pollution, and there is a lack of quantitative information regarding the microbial pollution characteristics of riverine water, despite frequently reported gastrointestinal diseases. The aim of our study was to apply faecal taxation (i.e., faecal pellet counting in representative test areas to estimate the potential availability of diffuse pollution sources) in combination with a detailed microbiological faecal pollution analysis in a riverine environment to elucidate the importance of diffuse pollution. To realize this approach, ambient faecal pellets, a multiparametric data set for standard faecal indicator bacteria (SFIB), including Escherichia coli, Clostridium perfringens spores and enterococci from catchment soil and river water, and a number of riverine water physicochemical variables were analysed during a one-year cycle. We demonstrated that the abundance of ambient faecal pellets, which were consistently counted at reference sites in the catchment, was associated with faecal pollution in the river water. Water SFIB, dissolved oxygen, nutrients, conductivity and total suspended solids were strongly linked with the abundance of ambient faecal pellets in the river catchment, as demonstrated by principal component analysis (PCA). Elevated concentrations of SFIB in the riverine water in the absence of rainfall also suggested the direct input of faecal bacteria into the riverine water by livestock (e.g., during watering) and humans (e.g., during bathing). Statistical analyses further revealed that the microbiological water quality of the investigated riverine water was not influenced by SFIB potentially occurring in the soil. This study demonstrates the importance of diffuse faecal pollution sources as major drivers of the microbiological quality of riverine water in the Ethiopian highlands. In addition, the new successfully applied integrated approach could be very useful for developing predictive models, which would aid in forecasting riverine microbiological quality in tropical developing countries.