For a specific range of building and occupancy types, this report examines the effectiveness of automatic sprinkler systems installed in New Zealand and Australia. The aim of the research is to quantify the likelihood of a fully developed fire occurring in sprinklered buildings. By deriving an annual probability of occurrence this can be compared with the accepted exceedance probabilities that exist for other limit state design actions for the design of steel structures. Comprehensive data collated for the entire history of sprinkler installations in New Zealand and Australia is analysed to obtain conditional probabilities that confirm the effectiveness of sprinklers to control fires. These probabilities correspond to the likelihood of fully developed fire occurring being classified as an extremely unlikely event. Passive fire protection is normally provided to protect a structure against a fully developed fire. It is therefore suggested that certain types of structural steel frames in sprinklered buildings do not require passive fire protection to meet performance requirements of the Building Code. The performance of steel frames without fire protection when exposed to fire following earthquake is assessed in a probabilistic framework. The likelihood of damage to the steel frame is not very different for the scenario of fire without earthquake. This report also examines other aspects that affect sprinkler reliability, such as town main water supply, system isolation due to internal alterations and booster pump reliability. Most of these items do not have a major effect on sprinkler reliability. Current literature describing the performance of steel framed multi-level buildings when subjected to fully developed fires is reviewed. The favourable behaviour of these real frames in natural fires confirms that the consequences are not usually serious if steel members without passive fire protection are exposed to severe fires.