Based on the Bayesian network approach and the Markov chain, this paper proposes models to forecast the propagation of Bacillus Necrosis Pangasius (BNP) disease on the shark catfish in a pond, ponds in the same area, and along river routes. The key factors considered in these models are: care and management conditions of fish farmers (such as water changes, nutrient levels, stocking densities); the environment change including physical factors (such as temperature, salty) and chemical factors (such as pH, H₂S, CO₂, heavy metals); the health status of fishes; and pathogens. Using these models, aquaculture experts can consolidate their judgments about the pathogenesis and the propagation of BNP disease, and especially can recommend fish farmers some ways to prevent BNP effectively.