Up-flow anaerobic sludge blanket (UASB) reactors are being used with increasing regularity all over the world, especially in India, for a variety of waste-water treatment operations. Consequently, there is a need to develop methodologies enabling one to determine UASB reactor performance, not only for designing more efficient UASB reactors but also for predicting the performance of existing reactors under various conditions of influent wastewater flows and characteristics. This work explores the feasibility of application of an artificial neural network-based model for simulating the performance of an existing UASB reactor. Accordingly, a neural network model was designed and trained to predict the steady-state performance of a UASB reactor treating high-strength (unrefined sugar based) wastewater. The model inputs were organic loading rate, hydraulic retention time, and influent bicarbonate alkalinity. The output variables were one or more of the following, effluent substrate concentration (Se), reactor bicarbonate alkalinity, reactor pH, reactor volatile fatty acid concentration, average gas production rate, and percent methane content of the gas. Training of the neural network model was achieved using a large amount of experimentally obtained reactor performance data from the reactor mentioned above as the training set. Training was followed by validation using independent sets of performance data obtained from the same UASB reactor. Subsequently, simulations were performed using the validated neural network model to determine the impact of changes in parameters like influent chemical oxygen demand (COD) concentration and hydraulic retention time on the reactor performance. Simulation results thus obtained were carefully analyzed based on qualitative understanding of UASB process and were found to provide important insights into key variables that were responsible for influencing the working of the UASB reactor under varying input conditions.