Precipitation estimation in areas without recorded meteorological data is of great importance in hydrological studies and flood forecasting. Due to the lack of high-altitude meteorological stations with long-term recorded data in Mazandaran province as well as stochastic attribute of rainfall data, using statistical methods based on covariates and comparison with geostatistical methods for interpolating monthly and annual rainfall data in this province is Inevitable. For this purpose, precipitation data from 21 meteorological stations over a 13-years recorded period (2004-2013) were used. In order to determination of appropriate interpolation method of rainfall data, six models including Ordinary Kriging, Cokriging, Inverse Distance Weighting, spline, three-dimensional linear gradient and regression-kriging were investigated. Evaluation of the methods was also performed on the basis of root mean square error, mean bias error and regression analysis. Variography analysis showed spherical and exponential models as the best theoretical semivariogram models. The results of error indices analysis showed the Spline model has the lowest efficiency and the three-dimensional linear gradient was found as the most appropriate interpolation model of rainfall data which in comparison with other models reduced the rainfall estimation error from 100 to 200 mm(about 40 to 60 percent). However, its accuracy is reduced in hot and humid months. Investication of rainfall maps illustrates the accuracy of covariate based interpolation methods in detecting low rainfall and high rainfall points of the province. So that high rainfall district is located on the west coast of the province and by moving east the amount of rainfall decrease. Due to the similarity of the rainfall map and the digital elevation model, it is noted that the amount of precipitation in valleys is very different from precipitation in the highlands of this province. The results of this study showed that in areas with complex topography, the use of covariates leads to a significant increase in the accuracy of rainfall maps.