Monitoring near‐surface soil water content is essential for efficient water management and for understanding hydrologic processes in soils. Ground‐penetrating radar (GPR) groundwaves are an approach that can be used to monitor the near‐surface soil water content, but the efficacy of this technique is currently limited by the uncertainty surrounding the groundwave sampling depth. This research experimentally determines the sampling depth of GPR groundwaves under dry and saturated conditions in a sandy soil. Data were acquired using 250, 500, and 1000 MHz antennas within an experimental tank containing soil layers of contrasting electromagnetic velocities. Results show that the groundwave sampling depth is a function of frequency in both dry and saturated soils, and sampling depth is inversely related to frequency. A comparison of data acquired under dry and saturated conditions indicates that the groundwave sampling depth is slightly less in saturated soil than in dry soil, but the dependence of sampling depth on soil water content may be less than has been predicted using numerical modeling. The minimum sampling depth observed in this experiment was 12 cm for the 1000 MHz antennas in saturated sand, and the maximum sampling depth was 30 cm for the 250 MHz antennas in dry sand.