Assessments of water repellency of soils are commonly made on air‐dried or oven‐dried samples, without considering the soil water content. The objectives of this study were to examine the spatial and temporal variability of soil water content, actual water repellency over short distances, and the variations in critical soil water contents. Between 22 Apr. and 23 Nov. 1999, numerous samples were collected from a grass‐covered dune sand (typic Psammaquent), at six depths, eight times in transects and two times in soil blocks. The water drop penetration time (WDPT) test was used to measure the actual water repellency of the field‐moist samples and the potential water repellency after drying the samples at 25, 65, and 105°C. Highly spatial and temporal variability in water content and persistence of actual water repellency was found between the samples from all soil depths. At each depth we established an upper water content, below which samples were water repellent and a lower water content, above which samples were wettable. This water content range, called the transition zone , was different for each depth, and, for example, assessed at 0‐ to 2.5‐cm depth between soil water contents of 18 and 23% (vol./vol.), and at 16.5‐ to 19‐cm depth between 2 and 5% (vol./vol.). The potential water repellency of samples dried at 25 and 65°C was on some days less severe than the actual repellency of field‐moist samples on other days, thus underestimating the maximal persistence of water repellency that can occur in the field. Drying of the samples at 105°C significantly increased the potential water repellency.