Model-based analysis is a well-established method to assess the dependability of a system before deployment. It is well known that, in highly dynamic contexts, the accuracy of the analysis results can be limited because unpredictable phenomena may aect the system during its operation. In such contexts, the analysis typically needs to be rened with data obtained from real system executions. In this paper we tackle the issue of rening model-based dependability analysis in automated systems through monitoring. Specically, we report on our preliminary results on the development of a system that exploits the synergic use of an automated approach for model-based dependability analysis and a exible monitoring architecture.