The aim of this article is to inform about research. which was done by use of artificial neural networks [ANN] applications for prognoses of Lithuania National Stock Exchange indexes LIT IN, LITIN-A, LITIN-VVP. Analysis for entropy shows the level of chance of tbe variations and correspondingly shows possibilities to find economic factors, which may influence Stock Exchange variations. Correlation analysis shows dependance between some Lithuania macroeconcmic indicators, foreign exchange indexes and LITIN, LITIN-A, LITIN-VVP indexes. It helps to include such indicators in to the autoregression, autoregression with the cause and cause prediction models. ANN learning is executed by weighted values of past period corresponding national indexes, by country’s macroeconomic indicators (like GNP, unemployment, inflation. interest rates and so on) and by other country’s Stock Exchange indexes (USA - Dow Jones and S&P, EU - Eurex, Russia - RTS). Comparison is made with the linear multidimensional regression method.