Global food production has grown faster than population in recent decades, due largely to improved seeds and increased use of fertilizer and irrigation. But crop yield growth is projected to slow, from more than 2 percent annually over the past four decades to about 1 percent annually over the next three—raising concerns that Malthus may yet prove right. Soil degradation's role in this decline has been widely studied but remains obscured by differences in methods, data, and scale of analysis. We reconcile these differences and find that erosion (the main form of soil degradation) reduces yields by 0.1–0.3 percent annually, on average, depending on farmers' incentives. Many soil science studies use plot-level experiments that (by design) lack data on the changing practices farmers employ in their fields. Many economic studies use data on production practices to examine productivity at broader scales but lack data on soil degradation. As a result, estimates of degradation's impact on yields vary widely—as low as 0.1 percent per year on a global scale and as high as 8 percent per year in the United States. This makes it difficult to assess impacts on food security or the environment, and thus the appropriate policy response. Recent research helps reveal degradation's impact by recognizing the roles of both nature and nurture—not in terms of tectonics and climate, as recently described, but biophysical processes and economic choices more generally. This helps us compare estimates from different approaches. A review of over 300 plot-level experiments around the world found that potential yield losses due to soil erosion vary widely by crop, soil, and region, but average 0.3 percent per year—assuming other factors do not change. But other factors do change, not least because farmers often have incentives to reduce erosion and its yield impacts—for example by reducing tillage or building terraces. Farmers' incentives depend on their access to information to evaluate alternative measures; savings or credit to finance profitable investments; and secure property rights to ensure long-run benefits will be realized. When these conditions are met (i.e. markets function well), optimal choices result in yield losses that are smaller than potential losses. Optimal losses under maize production in the United States are estimated at less than 0.1 percent per year. (Measures to reduce yield losses to zero are not optimal because they cost farmers more than they save.)When markets function poorly, farmers' incentives to adopt conservation practices are diminished. Second-best choices result in yield losses between potential and optimal levels (but we don't know exactly where, because they are masked by changes in seed quality and other factors). Assuming that actual yield trends reflect second-best losses to degradation averaging 0.2 percent per year, an estimated 37 million (5 percent) fewer people would be hungry in developing countries in 2010 if yield losses were reduced to optimal levels.Research incorporating farmers' incentives shows that soil degradation does not threaten productivity growth and food security at the global level, but problems remain in areas where resources are fragile and markets function poorly, particularly in developing countries. The challenge for policymakers is not to reduce erosion-induced losses from potential levels (which are unlikely) to zero (which is suboptimal), but rather from actual (second-best) levels to optimal. Priorities include enhancing farmers' incentives to reduce degradation by improving information, access to credit, and property rights. Addressing soil degradation's adverse effects beyond farmers' fields (e.g. climate change and national wealth) poses additional challenges.