The characteristics of, prediction models for, and experimental data pertaining to flashover in full-scale room fires were first reviewed. Then, initiation, growth, full development, and decay of three office fire scenarios were experimentally explored using a 10 MW fire test facility and continuous online combustion gas analysis. The conditions for flashover were investigated and compared with correlations in the literature. The model office compartment is an aerated lightweight concrete structure with dimensions of 5 m  6 m and with a net room height of 2.4-3.3 m. The results show that the measured minimum heat release rate at flashover is consistent with the correlations of Babrauskas [5] and McCaffrey et al. [6]. Based on the fundamental definition of flashover using the 'energy-filling' concept it is possible to predict the flashover time via a case-based reasoning method. However, more work is needed to further validate this concept.