To the insufficiency of the description for the fuzzy trend of uncertain data sets and the universe classification and to the problem of that the existing fuzzy time series forecasting theory is mostly limited to short-term realm, the model of long-term forecasting for intuitionistic fuzzy time series based on vector quantization, namely VQ-LIFTS is advanced. The classification of universe intervals is optimized with the intuitionistic fuzzy C-means clustering algorithm in the new model. Then the deterministic transition intuitionistic fuzzy rules are established by using back-tracking mechanism, and the non-matching historical patterns are solved efficiently by introducing sliding windows mechanism and vector technique. Therefore, the distribution characters of the uncertain time series data system are reflected accurately. The forecasting accuracy of long-term time series in the complex environment is also improved, thus greatly extending the fuzzy time series forecasting application. Finally, the experimental results validate the efficiency and preference of the proposed algorithm.