In many southeastern Coastal Plain soils, subsoil pans have strengths that restrict root growth. To reduce strengths, soils are deep tilled annually, and perhaps biannually for double cropping. We evaluated the effect of deep tillage in fall, in spring, or at both times on strength of a Goldsboro loamy sand (fine loamy, siliceous, thermic Aquic Kandiudult) and on the yield of wheat (Triticum aestivum L.) and drilled soybean [Glycine max (L.) Merr.] in a double‐cropped system. Treatments consisted of all combinations of surface tillage (disked and not disked) and deep tillage (no deep tillage, paratillage before wheat planting, before soybean planting, and before both) in four replicates. Soil strengths, measured as cone indices, showed that disked, non‐deep‐tilled treatments resulted in a pan at the 20‐ to 30‐cm depth, generally associated with an E horizon. In more recently and more frequently deep‐tilled treatments, mean profile cone indices were 0.31 to 0.36 MPa lower than treatments not deep tilled or deep tilled for the previous growing season. If soil was deep tilled only once a year, it was 0.26 MPa softer when tilled only in spring than when tilled only in fall. Deep tillage at the beginning of either season reduced soil cone indices and improved wheat and soybean yields over other treatments. Deep tillage at the beginning of both seasons maintained the softest soil. For every megapascal decrease in mean profile cone index, wheat yields increased 1.5 to 1.7 Mg ha−1 and soybean yields increased 1.1 to 1.8 Mg ha−1