The aim of the research is to control the quality of the pulp and the functional condition of the refiners using statistical process control (SPC) and fuzzy modelling techniques. The on-line measurements and the SPC parameters are combined with the Mamdani type fuzzy models. The models and the indexes are used for the optimisation and the fault diagnosis purposes in the industrial scale pulp mill. The methodology is applied to the reject refining line in the thermo mechanical pulp (TMP). The preliminary results are promising and the methodology can be applied to all type of TMP mills.