We apply two innovative econometric approaches to crop trials data to examine effects of rotations and fertilizer use on dynamics of soil quality and corn yields. First, we develop a random coefficients model of yield responses to nitrogen fertilizer and rotations to evaluate both short- and long-run substitutability of N fertilizer for rotation. Second, we construct a dynamic structural model to explicitly recover an indirect but general measure of soil quality. The results yield insights about optimal soil-conserving investments under asymmetric information.