The study uses the data collected from a household survey of 256 maize farmers in the Mekong Delta. Attitudes towards risk of maize farmers are measured by the Eckel Grossman's empirical method, using an experimental gambling approach with real payoffs. Risk aversion coefficients are determined by the utility function under the constant partial risk assumption (CPRA). The estimation results show that 46.48% of farmers are classified as extremely risk-averse; 21.88% as severely risk-averse; 13.28% as intermediate; 6.25% as moderate; 2.34% as slight to neutral toward risk and 9.77% as neutral to risk-loving farmers. In addition, this study also attempts to estimate economic efficiency and its determinants by jointly estimating the Cobb-Douglas stochastic profit frontier function and the inefficiency function. The estimation results show that the mean economic efficiency is 70.65%. Economic efficiency is found to be affected by farmers’ attitudes toward risk, education level of the household head, number of labours in the household, cultivated area, ratio of maize income to total household income. Among which, the higher risk-averse are the farmers the lower efficiency they achieve.