Traditional optimization methods have been applied for years to high-yield fertilization models, which are usually well formulated by crisp coefficients and variables. Unfortunately, real-world crop growing environment and process are often not deterministic. In this paper we establish a fuzzy mathematical model between
yield and fertilization application rates, in which variation coefficients of N, P, K are described with fuzzy numbers. In particular, we present a tabu search algorithm for finding a set of fertilization solutions in order to maximize
yield based on fuzzy measures including expected value, optimistic value and pessimistic value. Our approach is more realistic and practical for real-world problems by taking vague and imprecise data into consideration, provides more comprehensive decision support by generating a set of high-quality alternatives, and can be applied to fertilizer decision for a variety of other crops.