The article presents the possibility to control the real operation process of an arbitrary device installed in the marine power plant based on the four-state semi-Markov process, being the model of the process, which describes the transition process of operational states of the device (ek, k = 1, 2, 3, 4), and the transition process of its technical states (sl, l = 1, 2, 3). The operational states ek (k = 1, 2, 3, 4) have the following interpretation: e1 – active operation state resulting from the task performed by the device, e2 – state of ready-to-operate stop of the device, e3 – state of planned preventive service of the device, e4 – state of unplanned service of the device, forced by its damage. Whereas the interpretation of the technical states sl (l = 1, 2, 3) is as follows: s1 – state of full serviceability of the device, s2 – state of partial serviceability of the device, and s3 – state of unserviceability of the device. All these states are precisely defined for the ship main engine (SG). A hypothesis is proposed which justifies the use of this model to examine real state transitions in marine power plant device operation processes. The article shows the possibility to make operating decisions ensuring a rational course of the device operation process when the proposed model of this process and the dynamic programming method based on the Bellman’s principle of optimality are applied. The optimisation criterion adopted when making operating decisions is the expected profit to be gained as a result of functioning of the device in the time interval [τ0, τm], being the sum of the expected profit gained in interval [τ0, τ1] and to be gained in interval [τ1, τm].