The chaotic forecasting of irregular complex flow oscillation of natural circulation flow instabilities under rolling motion condition based on the largest Lyapunov exponents is performed. The correlation dimension, Kolmogorov entropy and the largest Lyapunov exponent are determined based on the phase space reconstruction theory of experimental data. On the premise that the irregular complex flow oscillation is confirmed to own chaos characteristic, the chaotic forecasting of the irregular complex flow oscillation is carried out by calculating the largest Lyapunov exponent. A comparisons between the prediction results and experimental data indicates that the chaotic forecasting based on the largest Lyapunov exponent is an effective way of producing those two-phase natural circulation flow instabilities. Meanwhile, the maximum predictable scale of chaotic flow instability is determined and a way of dynamic forecasting to monitor flow oscillation is presented. The method employed here provides a new method of studying the complex two-phase flow instabilities.