In this paper we apply the factor models to produce short-term forecasts for Lithuanian consumer and producer inflation. The factor models are compared with a random walk and the first order autoregression models. In this research work are used 147 time series publicly available at monthly frequency from 1996 until 2007. Research shows that, according Kaiser-Meyer-Olkin test, the most of these series (61 percent) are suitable for the factor analysis. The best forecast of producer price inflation is obtained by employing the factor model. The best forecast of consumer price inflation (according harmonised index of consumer prices) is made by employing the random walk model, but the factor model is better as the first
order autoregression model at short-term forecast horizon. Price data best describes consumer inflation for Lithuania and a single factor is needed to produce maximum accuracy of the model.