In this study, the procedures used in common methods for field measurement of unsaturated hydraulic conductivity in heterogeneous soil were simulated through numerical experiments. It was assumed that the heterogeneity of the field soil can be described by a scaling factor Δ which is treated as a lognormally distributed, second-order stationary random variable in space. After the hypothetical random field was generated using property variability and correlation length scales typical of those observed in field measurements, it was used to test various field measurement methods currently in practice. One detailed and three simplified unsteady drainage flux methods were used to estimate the hydraulic conductivity under plausible field measurement conditions. In addition, the soil water retention data were used to predict hydraulic conductivity from a pore-size distribution model. It was found that even under the assumption that there were no reading or instrument errors, other sources of errors can still be created due to the limitations of the observation techniques. These error sources include limited number of readings in space and time, separation distances between the readings of water content and pressure head, and volume-averaged readings rather than point estimates with common instrumentation. Under the conditions studied in the simulations, the detailed method of estimation produced reasonable estimates of hydraulic conductivity, but only when very dense measurements of water content and matric potential were used in the estimation. None of the approximate methods yielded accurate measurements of the true hydraulic conductivity of the field region measured. ¿ American Geophysical Union 1993