Soil water flux (J_w) is an important hydrological parameter which determines soil infiltration, runoff and solute transport, yet no corresponding practical techniques are available for its real-time measurement in situ. In this study, heat-pulse measurements were performed of deferent water fluxes in packed soil columns different in texture. Based on temperature changes between upstream and downstream locations from the heat-pulse probe, water flux was then calculated with the MDTD method (maximum dimensionless temperature increase difference), and the T_d /T_u method (ratio of temperature changes between downstream and upstream positions), and a modified T_d /T_u method along other known parameters. Results indicated that strong linear relationships ( R ²>0.99) existed between the estimated and the measured water fluxes (up to 6×10^-5 m s^-1). Though, the three methods all down-estimated, however, the T_d /T_u method was highest precise and simplest in calculation, and needed the least number of additional parameters. In coarse sand, the T_d /T_u method was relatively more accurate in estimation of J_w, yet in fine sandy clay loam, especially when J_w was greater, obvious discrepancies were observed between the estimated and the measured, reaching 20%. Besides, the reason of underestimation of J_w with the heat-pulse technique is explored and presented as a basis for further development of the heat-pulse technique.