Currently there is a need for a model to estimate mass emissions of atmospheric pollutants at the exit of the stacks of thermal power plants that operate under a variable regime of electric power generation based on the variables that typically are monitored during the operation of the plants. The recommended alternative to calculate the mass emissions of pollutants is based on the experimental measurements of pollutant concentration, velocity and temperature at the exit of the stack. This alternative is expensive and cumbersome to implement. Alternatively the US EPA emission factors can be used. However, the emission factors require modifications to account for the type of fuel, the technology used to control emissions, maintenance of the equipment, and the local environmental conditions. As a solution, this paper presents a model to estimate emissions of atmospheric pollutants in thermal power plants based on the variables that are continuously monitored during the operation of most of the thermal power plants in Mexico such as fuel chemical composition, fuel consumption, air to fuel ratio of the combustion process, and mean boiler temperature. The proposed model was calibrated by continuously measuring all the variables included in the three models during one week of operation of a 2.2 GW thermal power plant located in the continental area of the Gulf of Mexico. This plant has six units of generation that operate with fuel oil and one with natural gas. Results obtained from the three methodologies described before were compared. It was concluded that the NOx, SOx and CO results of the proposed model follow closely the results obtained using the measurements of concentration, velocity and temperature at the exit of the stack method. It was also found that the results of the emission factors methodology require to be adjusted to include the particular operating conditions of each unit of electricity generation.