Nowadays, there are many methods used to obtain mathematical models applied to active noise control, especially, when the transfer function is required. Inside rooms, the global active sound control has bad results due to the reflections and the diffuse field. Then, authors have applied system identification to find more complex mathematical models based on measured data. Also, the number of system identification methodologies is increasing and it carries to having many unexplored models. In order to know which models are useful for global active noise control inside rooms, a relationship between the sound pressure level decreased and the error of the mathematical model is presented. First, the concept of "a useful mathematical model" is defined under any context based on an analysis of the error (FIT). In addition, this concept is delimited to the active noise control context. Finally, an example is presented.