There are implicit assumptions that should be considered for the commonly used production-decline models (Exponential, Harmonic and Hyperbolic decline models), such as relative stable well control conditions, leads to the difficulty on quantitative forecasting during some complicated situation happens. A new decline prediction model was proposed through theoretical derivation, and the decline rate was interpreted from three different parts (water cut rising, liquid productivity index changing , and pressure drop changing ) for the first time. Furthermore, three calculation cases under different conditions (under constant liquid production rate, constant pressure drop and arbitrary conditions) were analyzed, and the different changing rules of deline rate were analyzed. Comparing to the actual dynamic data, the proposed model shows high accuracy, even if some complicated situation happens, however theprediction error of the common decline models could be large. So it makes great sense in improving the dynamic performance forecasting.