Fire occurrences in a particular building are really rare events. It is assumed that fires occur in accordance with the Poisson process and the number of fire occurrences in time interval can be modeled by a Poisson distribution. Using data such as the numbers of annual fire occurrences and building floor area the probability of fire occurrence in different occupancies can be estimated. In addition, the relation between the numbers of fire occurrence and the time of fire occurrence are clearly discussed. Investigations for different groups of building occupancy are illustrated in this research. Based on mean fire ignition rates and floor areas of different occupancies results show the ranking order regarding annual fire ignition per unit floor area among these occupancies. Industrial occupancy has the highest value of annual fire ignition rate, followed by residential occupancy. The rate of fire ignition only increases in public occupancy in Taiwan.