The subject of this research is various types of disturbances rejection acting on a continuous billet-heating furnace. This problem is solved with the help of a PI-controller online adjustment parameters. The task is to determine which parameter of this controller is to be tuned in order to reject a disturbance of a certain type effectively. In the furtherance of that goal, a number of subtasks are solved. First of all, a zone of the continuous billet-heating furnace is identified using steady state statistics. Secondly, disturbances are classified into two categories according to their cite of action at a control system mathematical model. The first disturbance «heating furnace charging with cold metal» is implemented as the plant model gain decrease, whereas the second disturbance «furnace opening to charge metal» is implemented as a sine signal without its negative part. Thirdly, both these disturbances are tried to be rejected by tuning different parameters of the PI -controller. Both KP and KI should be tuned to reject the disturbance of the first type (and mixed disturbance including both types at the same time)while it is enough to tune KP only for the second type of them.