With the introduction into EU necessary there is an accelerated improvement of insurance process in Latvia in particular development of fire insurance of residential buildings. This kind of insurance should be accessible to all inhabitants of Latvia irrespective of their social status. One of important instruments of social protection is compulsory insurance. By applying it the state provides an indemnity to those persons who suffered from the occurrence of certain eventualities. The major advantage of compulsory insurance is the big volume of insurance area that reduces the hazard of negative risk selection. In the authors’ opinion, one of the key drawbacks of the present system of fire insurance, which prevents introduction of compulsory fire insurance, is absence of a complex unified system of the assessment of the level of fire risk at the objects, and methods of insurance premium calculation. In the development of the concept of compulsory fire insurance, it is necessary to calculate possible insurance premiums which are connected with this kind of insurance depending on a kind of residential buildings. The application of imitation modelling is connected with the fact that frequently it is not possible to provide a definite description of the behavior of the insurance process being investigated. In paper at the modelling initial stage it is considered three types of inhabited constructions: wooden, stone and mixed. Depending on the sizes of insurance payments the question of participation of the state in this insurance process can be solved with purpose in full or in part to give indemnification of payment of awards to those inhabitants of Latvia who are needy. The basic attention in work is given to construct imitating model of insurance process allowing receiving exact enough estimations for size of premium for a various kind of residential buildings. In work by means of a statistical method of modeling, two major components of insurance premium have been calculated: net premium and risk's premium. Calculating other components of premium can be done using well known actuarial procedures of calculations. For modelling such complicated process as is fire insurance process it is Important to find more appropriate modelling method. For this purpose authors in paper compared two methods of modelling nonparametric and parametric. Nonparametric method doesn’t use information about insurance process parameters and can be modeled using only empirical histograms – marginal empirical distributions of each incidental value (see Figure 4). In most cases when there is not enough information for exact calculation of parameters of distribution, the most effective is application of a nonparametric method of modelling for research of real process because the available information on occurring process is used. As result in paper are presented confidence intervals for net and risk premiums for three types of residential buildings.