In this study we designed and tested a methodology to minimize the variation of soil matric potential due to changes in soil water content, thus achieving a close control over the water regime in a microcosm, used to study the biocontrol of Pythium sp. infection of seeds by antagonistic bacteria. The variation of volumetric water content and matric potential were monitored at different depths during an average experimental period of 14 days in soil, contained in replicate 1 315 cm3 microcosms, each sown with 16 sugar beet (Beta vulgaris) seeds and placed in a phytotron. Several experiments with target soil matric potentials of -10, -100 and -300 kPa were performed. It was found possible to maintain a matric potential of about -10 kPa in soil with minimum water content gradients between different parts of the microcosm by watering uniformly at each depth. With an appropriate watering and sampling procedure it was also possible to control m between -150 and -50 kPa, with a target value around -100 kPa; and to control m in the range between -250 and -375 kPa for a target value of -300 kPa. In all experiments spatial and temporal variability of matric potential was larger at the top than in the middle and in the bottom of microcosms. Variability in matric potential between replicate microcosms within each experiments was attributed to slight variation in packing, soil texture and drying rate, initial VWC.