Stiffened panel is a fundamental member of hull structures which is designed to withstand multiple loads such as revered cyclic load and lateral pressure. At present the ultimate strength of stiffened panel is conventionally determined by the monotonic compression criterion, ignoring the real progressive collapse of ship in rough sea conditions. To provide useful insights for improving the ship design, the present study focuses on dealing with the structural collapse characteristics and ultimate strength estimation of continuous stiffened panel under combined longitudinal extreme cyclic load and lateral pressure. A large number of nonlinear finite element analyses (FEA) are conducted to identify the influences of various factors on the failure mode and ultimate strength including the slenderness ratio of local panel, stiffener type, size and number, sequence and amplitude of the cyclic load, cycle number and lateral pressure. It is found that the ultimate limit state behaviours are governed by the coupling effects of several influential parameters. Based on artificial neural networks (ANN), a prediction model with high accuracy is proposed to evaluate the ultimate strength in different cycles.