The paper presents semi-Markov models of technical state transitions for diesel engines, useful for determining the reliability of engines. A possibility of application of a three-state model with a simplified matrix function, or even a two-state model, to determine reliability of the engines, has been described herein on examples of known from literature semi-Markov models, for the case when appropriate diagnosing systems (SDG) are used to identify the technical condition of the engines considered as diagnosed systems (SDN). A risk function and a renewal (restitution) function have been proposed for developing a two-state model of engine state transitions. An opportunity for considering the reliability of diagnosis for making operational decisions by applying the statistical decision theory, has also been presented. Conditional probability P(si/Ki) is recognized as a measure of reliability of diagnosis