Using nationally representative, multiyear farm-level data from the 1998–2023 Agricultural Resource Management Survey (ARMS), this study investigates long-run patterns and determinants of crop diversification in U.S. agriculture. Fractional response logit models estimated for four diversification indicators—the largest crop share, top-two crop share, Herfindahl index, and Shannon index—show that larger farms and mixed crop–livestock operations exhibit higher levels of diversification. The analysis further demonstrates that land ownership, land capability, operator characteristics, and household dependence on farm income significantly influence diversification choices. Interactions among land ownership, farm size, and land capability modify the magnitude and direction of these effects. Warmer growing conditions are associated with greater diversification, whereas extreme heat leads to increased crop concentration. Substantial regional variation in diversification is also observed.
