Measurements of sorptivity are reported for a number of field soils. The main method, used in situ, is simple and consists of timing initial vertical flow of water into an undisturbed soil profile. This method relies on the assumption that, during the measurement time of 1-2 min, the first term of the infiltration equation i = St1/2 + At + Bt3/2 . . . dominates flow. In other methods horizontal inflow was timed into undisturbed soil blocks, and capillary rise was observed in undisturbed field cores. Sorptivity was quite variable on each site, with standard deviations of the same order as those found for hydraulic conductivity. For most soils the three methods estimated sorptivity differently, the field method giving the largest values. These differences were largely due to the effect of water depth over the soil on sorptivity. Head dependence of sorptivity, verified experimentally and with delta function theory, was particularly strong in soils of composite structure, where infiltration is dominated by structural cracks and root channels. These soils also have relatively large values of the ratio of hydraulic conductivity to sorptivity. Large ratios of hydraulic conductivity to sorptivity and strong head dependence of sorptivity may cause significant errors in calculating sorptivity in the field method. However, these errors are compensating, and analysis of the data indicates that over-estimation of sorptivity in such soils is at the most 5%. For subsoils and soils of single grain structure these errors were negligible. Average sorptivity values of some of these soils have given good quantitative agreement between theory and field data on two-dimensional flow.