The ultimate goal of the research presented in this paper is analysis, modeling, and prediction of failure in an embedded system. To this end, we propose software instrumentation that can facilitate monitoring and evaluation of the status of an embedded system. We illustrate the approach with a simple case study and describe how population of a failure model can be carried out using the data collected. The information gained facilitates assurance of embedded systems, as it provides a means for collecting the data required for determining dependability attributes such as reliability and survivability.