Reactive power planning involves numerous factors of uncertainty. The application of the multi-scenario technique and fuzzy technique in the power system uncertainty planning is analyzed. Owing to the fact that either technique is only suitable for the single type of uncertain factors， a comprehensive reactive power planning model is presented based on the concept and advantages of combining the two techniques. The model takes the multi-scenario technique and fuzzy technique to correspond to such different types of uncertainty as network topology and load forecasting， respectively. The expected reward of the reactive equipment investment is used as the objective function to make the target scheme probabilistically the optimal one. A multi-population genetic algorithm is presented to implement the model. With the reactive power planning of an actual sub-transmission network as an object， the model and the algorithm have proved effective and superior.