The ultimate aim of this paper is to generate micro-level risk measures, which can provide a useful input for further research. To this end, this paper estimates bankruptcy probabilities for Hungarian firms using probit estimation. The estimated models show reasonable performance in distinguishing surviving and failing firms. We combine macro and micro information, as the addition of macro variables is needed to capture the aggregate dynamics and level of risk, especially during the crisis period. Controlling for the non-linear impact of firm characteristics and allowing heterogeneity by firm size improves the model's performance significantly. The distributional characteristics of the micro-level risk indicators provide some interesting insights regarding the development of risk dispersion and the risktaking of the banking sector.