This paper examines the issues encountered in the modelling dynamics sales of apartments during a period of realization of new building. Point is that a forecast of cash flow (income) is very important question for developers and investors. How much apartments we would sell in first month? In second one? When the last apartment will be sold? An extensive listing database of the apartments for sale has been employed to build the model describing the realization process. This database consists of about 1200 addresses of new buildings sold in 2002-2006 years in Moscow city and region. We have analyzed the dynamics of sales for these buildings to found the most suitable formulae for modelling this process. It seems that the best curves family is the so-called logistic curves, describing start, movement, and braking of a car. Of course we need to improve and modify logistic curve to reach the best fitting of the real situation. No doubt, the model does not purport to explain everything about real process on the apartment market. It has a lot of limitations: competitors on the market, changing in prices during realization, etc. But, at least, the model allows us to approximately predict the income in dependence of time. The last chapter of the paper is dedicated to the specificity of Moscow housing projects. With the help of our model we will try to explain the current strategy of new apartment sellers in Moscow city, psychology and behaviour of buyers.