Two-factorial experiment was conducted on loess soil in Czesławice Experimental Station in 1997-1999 by split-plot method in four replications. Two soybean cultivars – Aldana and Polan and five methods of soil tillage were of experimental factors. Tillage treatments included: I. conventional tillage, II. reduced tillage – without stubble operations, III. reduced to cultivator and harrow and then subsoiler application, V. reduced tillage – direct sowing of white mustard, harvest of mustard in late autumn and then deep mouldboard plough was performed, V. reduced tillage – direct sowing of white mustard then mustard was sprayed by Reglone (diquat) in spring, and finally soybean was sown by direct sowing method. Results revealed that the highest water store in every sampling time and under each cultivar was kept by soil under IV tillage treatment, less – under conventional tillage, I. while least one – under II and Ill tillage treatments. Amongst reduced tillage technologies of soil cultivation the water was stored in the soil quite well also when of V. tillage method was applied. Under Aldana cultivar soil was significantly dried than under Polan cv. Soil bulk density and total and capillary porosity did not showed larger variability in dependence on treatments applied.