In the present paper, we aim at providing plug-in-type empirical estimators that enable us to quantify the contribution of each operational or/and non-functioning state to the failures of a system described by a semi-Markov model. In the discrete time and finite state space semi-Markov framework, we study different conditional versions of an important reliability measure for random repairable systems, the failure occurrence rate based on counting processes. The aforementioned estimators are characterised by appealing asymptotic properties such as strong consistency and asymptotic normality. We further obtain detailed analytical expressions for the covariance matrices of the random vectors describing the conditional failure occurrence rates. As particular cases we present the failure occurrence rates for hidden (semi-) Markov models. We illustrate our results by means of an academic example based on simulated data. An application to real data is further presented that models sustainable vibration levels in a semi-Markov framework.Semi-Markov models (SMMs) are state-of-the-art models that are widely used in many scientific fields such as reliability and DNA analysis ([3]), seismology ([22]) etc. One of the main distinguising features of SMMs is that contrary to Markov models, they enable us to describe systems that evolve based not only on their last visited state (Markov property) but also on the time elapsed since this state. Due to this feature, popular "memory-full" distributions, such as the Weibull distribution, could be employed to describe sojourn (or interevent) times between successive events. We refer the interested reader to [8] or [12] for an introduction to homogeneous SMMs and to [18,17] for non-homogeneous SMMs, respectively.In the semi-Markov context, many reliability indicators have been introduced, including mean times to failure, hazard rates, availability functions etc. For recent advances in the topic concerning discrete time SMMs, see [1], [3], [5] and [6]. For continuous time SMMs, we address the interested reader to [10] and [11]. For advances in estimation methods of nonparametric semi-Markov models, see [15] and the references therein.Here we focus on the rate of occurrence of failures (ROCOF), which is a fundamental reliability indicator for repairable, random systems subject to multiple failures. Yeh ([23]) was