Probabilistic assessment methods can be used to identify specific plant vulnerabilities. Application of such methods can also facilitate selection among system design alternatives available for safety enhancements. The quality of assessment results is however strongly dependent on realistic and accurate input data for modelling of system component behaviour and failure modes during conditions to be assessed. Use of conservative input data may not lead to results providing guidance on safety upgrades. Adequate input data for probabilistic assessments seems to be lacking for at least failure modes of some electrical components when exposed to a fire. This report presents an attempt to improve the situation with respect to such input data. In order to take advantage of information in existing documentation of fire incident occurrences some of the lessons learned from the fire at Browns Ferry Nuclear Power Plant on March 22, 1975 are discussed in this report. Also a summary of results from ...