In view of the problem that support vector regression(SVR)cannot provide better porosity prediction because different lithologic reservoirs have different pore types and different porosity structures, a new model for estimating porosity, taking lithology information into account, is proposed. In the model, the lithology information of the sample is converted to attribute values that are closely associated with the lithology information. A method that combines the grid search algorithm for rough screening and intelligent search algorithms(genetic algorithms and particle swarm optimization)for fine filtering was used to optimize the model parameters. The grid search algorithm for rough screening was used to determine the approximate scope of the optimal solution, and intelligent search algorithms for fine filtering were used to determine the optimal solution in a local region. The optimized parameters were used to establish the forecasting model. The predicted results were compared with the measured data. The results show that the prediction accuracy of the model is greatly improved when the lithology information is taken into account, and the prediction accuracy of the intelligent search algorithms for fine filtering is higher than that of traditional methods.