Operational and reliability problems may arise after photovoltaic system grid integration,due to PV power’s characteristics, such as volatility and randomness．Predicting the photovoltaic power accurately in advance will benefit the power sector to adjust the scheduling plan timely to ensure power quality, and then ensure the grid operate safely. Based on the analysis of the limitations of the gray forecasting model, the Gray - Markov chain combination forecasting model was applied to predict the short-term photovoltaic power ,and described its modeling principles. Gray - Markov chain model can be drawn through analyzing the relative residuals of the fitted values ??of the gray model and establishing the state transition probability matrix. Through analyzing the practical examples, verified the accuracy and simplicity of the proposed combination model. #$NLKeywords: photovoltaic power prediction; grey model; Markov chain; short-term forecasting