Specific height dependences of the precipitation based on the daily meteodata analysis (for the period 1970-85) were obtained for the Western Sayan, that is under conditions of insufficient data provided by the scarce network of stations. The analysis revealed that annual precipitation sum is characterized by closer dependence rather than seasonal one. It has been generally observed that the linear correlation has the highest reliability for southern macroslope of the ridge. At the same time the dependence of northern slope is primarily determined by the height: distribution of precipitation is defined by logarithmic function. Thus, low mountains pluviometric gradient (below 500 m) of warm period reaches a value 100 mm/100 m. As for the mid- mountain parts (above 1300 m), gradients of warm and cold periods don't exceed 2 mm/100 m. Reliability of identified dependences has been confirmed by model calculations. All the modelled objects and periods were selected considering the existence and availability of both hydrological and meteorological data - the river Us and Amyl basins (right tributaries of Yenisei, Krasnoyarsk district, Russia) which represent typical runoff conditions for the semi-arid and semi-humid areas of the West Sayan. The investigation also found that the application of seasonal pluviometric gradients during the runoff modeling has improved the convergence of calculated and obtained hydrographs. It was shown that taking into account vertical pluviometric gradients turned out to be more significant for semi-arid regions. And furthermore, it's improved the quality of runoff modeling notably for periods, characterized by the least value of Nash-Sutcliffe criteria, and reduced an amount of years with improper quality of calculation. This is especially important under conditions of insufficient hydrometeorological data.