The methodology of building of discrete dynamic models aimed at describing of the processes of drug addiction propagation, general assessing of the drug situation on the territories, revealing of the role of different factors in its formation and short-term forecast of its development. Two types of different models are considered: quadratic stationary and linear non-stationary models, which are accurately identified. The methodology of identifying of models and calculating of the lower estimate of the number of drug addicts and the number of drug pushers on the territory during the current period is described. The two models are compared, and the linear model is recommended for practical use in state organisation. Real examples show that the application of the methods of smoothing of time series and calculating of assessments along the trend which is constantly specified according to the new data obtained enables to increase the length of the forecast interval. Accurate assessment of the existing reality according to the results of low-enforcement and medical establishments ’ work suffices for the calculation; it is recommended that expert estimates of respective specialists are used.