Abstract Growers and scientists have observed that soybean ( Glycine max (L.) Merrill) yield decreases after the second or third year of a continuous soybean no-tillage system. In 1984, at the end of an 8-year double-cropping oat ( Avena sativa L.)/soybean tillage experiment, soil compaction and its association with soybean yield were evaluated. The soil series was an Arredondo fine sand (loamy, siliceous, hyperthermic Grossarenic Paleudulte). Treatments were no-tillage (NT), no-tillage plus in-row subsoiling (NTPS), conventional tillage (CT), and conventional tillage plus in-row subsoiling (CTPS). Forty days after planting soybeans, soil penetrometer resistance (SPR) readings were taken to a depth of 60 cm at 5 positions; in the row and at 15 and 30 cm from the row on both traffic and no-traffic sides of 4 rows. Row spacing was 76 cm. Interaction existed among treatments, positions and depth of sampling. For the no-traffic inter-row at the 10-cm depth, the mean SPR readings for the 15- and 30-cm positions were 0.4 and 1.6 MPa, respectively, for the CT and NT treatments. Maximum SPR (3.2 MPa) occurred at the 30-cm depth for all treatments (CT, NT, CTPS, NTPS) and inter-rows (traffic, non-traffic). In-row subsoiling at planting, reduced SPR to less than 0.4 MPa in the top 20 cm, but compressed the soil vertically below 35 cm and at this depth laterally as far as 30 cm. Soybean yields were lowest, but similar (1.34 and 1.20 Mg ha −1), for CT and NT. In-row subsoiling increased yields 35 and 48%, respectively, for CT and NT. The lowest SPR values (1.57 and 1.59 MPa) were associated with the highest yields (1.82 and 1.78 Mg ha −1), for CTPS and NTPS. The increase in yield for subsoiling (NTPS vs. NT) amounts to about 14 000 MJ ha −1 for an energy input cost of about 820 Mj ha −1.