Prior studies have demonstrated the importance of maintenance in raising mining equipment performance levels together with pollution prevention and cleaner production. However, operators’ working conditions have not been studied enough in the previous research. In that aim, in this paper, firstly, operators’ working conditions in 10 excavators working in Serbia have been analyzed. Later on, the study uses artificial neural networks (ANN) for developing a quantitative model for estimating the failure rate of excavators. In order to avoid potential indirect financial losses, which sometimes exceed 15,000 euros per hour, the duration times of 590 excavator downtimes, measured over 198 days at the Serbian mining sites, were used as an input to the ANN. This enables the classification of failures lasting more than an hour based on the preceding 14 days. The most common type of downtime was found to be technological, according to a Pareto analysis of the observed data. The findings demonstrate that the non-linear link between excavation operations and excavator failure rates could be mapped using ANN modeling. The results also showed that operators’ working conditions very often have exceeded the boundaries prescribed by regulation, and the suggested ANN model offers a precise estimate tool for predicting excavator failure rates during the planning stage. The future research avenue is to continue monitoring working conditions and failures and to predict in real time precisely the length of time that an excavator would be down and to find deeper interrelations between operators’ working conditions and an excavator’s downtime.