This paper proposes an alternative measure of irrigation water efficiency based on the concept of input-specific technical efficiency. It uses a stochastic production frontier approach, based on Battese and Coelli’s (1995) inefficiency effect model, to obtain farm-specific estimates of technical and irrigation water efficiency, and a second-stage regression approach to identify the factors that influence irrigation water efficiency differentials. This methodology was applied to a sample of 144 citrus farms in Nabeul, Tunisia. Technical efficiency varies widely, suggesting that these citrus farmers could increase their production by as much 33% by using inputs more efficiently. Mean irrigation water efficiency is similarly variable, and on average below technical efficiency, suggesting that they could produce the same quantity of citrus using the same quantity of inputs but 47% less water. Finally, the results showed that the farmer’s age, education level and agricultural training, and the farm’s size, share of productive trees and availability of water tend to affect the degree of both technical and irrigation water efficiency positively.