Since sulfur release and capture processes during solid fuel combustion in circulating fluidized bed (CFB), especially in oxy-fuel conditions are very complex, the development of a simple SO2 emissions model for a wide range of operating conditions both for large- and pilot-scale boilers is of practical significance. Previously established and validated [16-1-6-1] ANN model, which was published in the Part 1 of this paper was employed to predict SO2 emissions from coal combustion in a large-scale 261 MWe, CFB COMPACT-type boiler as well as in a pilot-scale 0.1 MWth OxyFuel-CFB test rig. The simulations are carried out using artificial neural network approach for different combustion environments, both in atmospheric and pressurized conditions. The study is conducted for air-firing, oxygen-enriched and oxy-fired combustion conditions. Therefore, four different combustion atmospheres are considered in the paper, where combustion runs in air and air enriched with oxygen (O-2/N-2 mode) as well as in oxycombustion (oxygen-fired combustion) conditions, which mean the mixture of oxygen with CO2 or recycled flue gas (RFG) with various fractions of oxygen (O-2/CO2 mode and O-2/AFG mode, respectively). The obtained results show that the ANN model makes it possible to predict the SO2 emissions from coal combustion in CFB boilers of different sizes and in different combustion environments. (C) 2015 Elsevier B.V. All rights reserved.