Modeling crop yield distributions has been an important topic in agricultural production and risk analysis, and nonparametric methods have gained attention for their flexibility in describing the shapes of yield density functions. In this article, we apply a nonparametric method to model joint yield distributions based on farm-level data for multiple crops, and also provide a way of simulation for univariate and bivariate distributions. The results show that the nonparametric models, both univariate and bivariate, are estimated quite well compared to the original samples, and the simulated empirical distributions also preserve the attributes of the original samples at a reasonable level. This article provides a feasible way of using multivariate nonparametric methods in further risk and insurance analysis.