Actual models of water flow in soil do not take into account a heterogeneous structure of soil. However water flow in soil has non-uniform character, both in space, and in time. The quantitative estimation of these processes and construction of adequate models of preferential flow is very important for environment. The aim of tests is to observe microscale heterogeneity of infiltration and flow in different soils and compare results, to estimate distributing of quantity of infil-tration water and flow on area. The infiltration tests were carried out from July 2002 till August 2005 in irri-gated fields of the Saratov region (Volga region) and the Rostov region of Russia. Experiences were carried out on black-chestnut and chernozem soils. For the differentiated gathering the filtered water we used microlysimeter, consisting of 36 individually sampled cells of 6.7 by 6.7 cm. Each cell was connected to the corresponding graduated receiver for measurement of flow volume. Also there was used microlysimeter, consisting of 64 cells for tests in 2005. Monitoring of moisture was carried out with the probe TDR. For this type of measurements in the middles of monoliths were made boreholes in which were installed a tube for soil moisture scanning. For reduction of influence of a lateral filtration the water-ing area exceeded the microlysimeter area in 4 times. For the estimation of hetero-geneity infiltration on a surface of soil there was installed the frame, consisting 9 sections. Microlysimeters were installed on the following depths: 30, 50, 55 and 80 cm. Watering rate varied from 150 up to 1900 m3/ga. Watering were carried out by two ways: maintenance of the certain layer of water on a monolith and watering in the portions. The tests showed, that heterogeneity of infiltration is enough expressed on both types of soil. The maximal height of infiltration water layer in the separate section can exceed mean value of layer on all sections in 3 and more times. Flow distribution is very heterogeneous. Sometimes maximal flow through the separate cells is in 10 and more times higher then mean value of all cells. In some cells we did not get a flow even when moisture was very high (> 40%). While a plenty of flow passes through 5-7 cells for black-chestnut soil and 3-5 cells for chernozem soil. We can observe high correlation between contiguous steps of watering both in infiltration and in flow. With increasing moisture of soil the correlation coefficients also increasing and its value became closer to 1. For chernozem soil the correlation coefficient is higher than for black-chestnut soil.