The article discusses damages of essential tribological associations in crankshaft and piston systems of large power two-stroke engines used as main engines, which take place during transport tasks performed by those ships. Difficulties are named which make preventing those damages impossible, despite the fact that the technical state of engines of this type is identified with the aid of complex diagnostic systems making use of advanced computer technology. It is demonstrated that one of causes of the damages is the lack of research activities oriented on recognising random properties of the loads leading to those damages. A proposal is made for the loads acting at a given time t on tribological associations in crankshaft and piston systems of internal combustion engines used as main engines to be considered as random variables Q t . At the same time the loads examined within a given time interval t r ≤ t ≤ t z would be considered stochastic processes {Q(t): t ≥ 0}. Essential properties of the loads of the abovementioned tribological associations are named and explained by formulating hypotheses which need empirical verification. Interval estimation is proposed for estimating the expected value E(Q t ) of the load Q t acting at time t. A relation is indicated between the mechanical load and the thermal load acting on tribological associations in the ship main engine crankshaft and piston system.A suggestion is formulated that a stochastic form of the relation between these types of load is to be searched for, rather than statistic relation, and a proposal is made to measure the intensity (strength) of the stochastic relation using the Czuprow's convergence coefficient.