The general theme of this dissertation is agricultural risk and insurance in the United States. Chapter 2 examines welfare effects of the 2002 farm bill programs and yield insurance as well as their impacts on acreage decision of a representative Iowa farmer. Instead of measuring welfare using expected utility to capture farmers' preferences over risky alternatives, we apply recent advances in decision theory and use prospect theory to measure welfare changes due to government programs. The results indicate that there is no policy distortion to farmers' acreage decisions and farmers' willingness to pay per dollar of program cost is greatest for crop insurance. Given that farmers have crop insurance, the willingness to pay per dollar of program cost is much lower for loan deficiency payments, direct payments, and counter-cyclical payments. Chapter 3 develops a method for determining the aggregate risk of a book of business using hail insurance data. A spatial statistical approach is employed to measure the spatial correlation of hail loss cost. Monte Carlo simulation techniques are employed to simulate hail losses for a wide range of books of business. A regression model is estimated that captures the essence of the Monte Carlo simulation. This model can then be used to quickly estimate the degree of poolability of any given book of business. Chapter 4 turns to weather-based index contracts as alternative risk-management instruments in agriculture. A major concern associated with index contracts is basis risk. To address spatial basis risk, two spatial interpolation approaches, a geo-statistical approach and a Markov random field approach, are compared. The Markov random field approach is preferred because it has a smaller cross-validation prediction mean squared error. A temperature index insurance is presented based on interpolated data. The potential performance of the proposed index insurance is investigated through historical analysis in contract years 1980 to 2005.