There are concerns that restored grasslands currently under the Conservation Reserve Program (CRP) will experience increased soil erosion when they are returned to crop production. Our objective was to compare runoff, erosion, and erodibility on CRP land converted to annual hay production (permanent hayed, PH) and crop production under conventional-till (preplant disk tillage, CT) and no-till (NT) management. Erosion study was conducted in central North Dakota on Typic Argiustoll soil 6 yr after the CRP land had been converted to hay production and crop production with a spring wheat (Triticum aestivum L.) –winter wheat–dry pea (Pisum sativum L.) rotation. Runoff volumes and soil loss (by alum precipitation of sediment) were measured on 1.5 by 5 m bordered plots on 4% slope under a rainfall simulator delivering 1- to 3-h rains at 50 or 75 mm h−1, followed by three rains of 20 min or less at rates from 25 to 125 mm h−1 Erodibility was calculated from ratios and regressions of soil loss rates versus runoff rates measured at relative steady state. Runoff rates from 50 and 75 mm h−1 rains for CT, NT, and PH averaged 9, 12, and 21 mm h−1, respectively, and supported soil loss rates of 20, 7, and 8 g m−2 h−1 Erodibility of undisturbed CT, NT, and PH was 1.65, 0.29, and 0.28 g m−2 mm−1, respectively, showing NT did not differ from PH and that CT management increased erodibility six-fold above PH. Thorough disk tillage increased erodibility three-fold over CT, 15-fold over NT, and nine-fold over PH. Complete, nondisturbing residue removal increased erodibility less than tillage, from 1.2 times for CT to 2.5 times for NT. Chemically weeded NT exhibited the same low erodibility as the grassland PH treatment under the conditions of study—4% land slope, above average precipitation, and a residue-productive crop rotation. However, erodibility of tilled NT was significantly higher than that of tilled PH, showing the higher inherent stability of grassland surface soil with its perennial plant root structures.