Many long railway tunnels without emergency evacuation system or ventilation system are under construction or in-use in Korea. In the case of tunnel-fire, many fatalities are occur in current condition. Current safety level is estimated in this study, for the efficient investment on safety. But so many uncertainties in major input parameters make the safety estimation difficult. In this study, probabilistic techniques are applied for the consideration of uncertainties in major input parameters. As results of this study, accident scenarios and survival ratio under tunnel fire accident are determined for various conditions.