Estimates of groundwater volumes available in semiarid regions that rely on water balance calculations require the determination of both surface to groundwater lag times and volumes from irrigation or rainfall initiated recharge. Subsurface geologic material hydraulic properties (e.g. hydraulic conductivities, water retention functions) necessary for unsaturated flow modelling are rarely available as are the instrumented field tests that might determine such lag times. Here we develop a simple two-parameter (specific yield, Sy, and pore-size distribution index, λ), one-dimensional unsaturated flow model from simplifications of the Richards equation (using the Brooks-Corey relationships) to determine lag times from agricultural deep drainage associated with the irrigation of alfalfa hay and various row crops in the Antelope Valley of California, USA. Model-predicted lag times to depths of 85 m bgs (below ground surface) were similar to that measured in a 2-year ponded recharge field trial, slightly overestimating that measured by approximately 15% (0.51 vs 0.44 years). Lag time estimates were most sensitive to estimated deep percolation rates and roughly equally sensitive to the model hydraulic parameters. Generally, as subsurface material textures coarsen towards larger Sy and λ values for all Sy >10%, lag times progressively increase; however, at Sy <10%, lag times decrease substantially suggesting that particular combinations of Sy and λ values that may be associated with similarly textured materials can result in the prediction of different lag times for Sy approximately 10%. Overall, lag times of 1–3 years to a depth of 69 m bgs were estimated from deep drainage of agricultural irrigation across a variety of irrigation schedules and subsurface materials. Copyright © 2012 John Wiley & Sons, Ltd.