This paper proposes a new technique for designing fractional order controllers applied to the automation of main canal pools of a water irrigation system. As large variation of all the plant parameters is present in such systems, a fractional order family of controllers combined with a Smith Predictor is designed using time domain specifications. The designed controllers are compared with the standard combination of a Smith Predictor with a proportional-integral (PI) controller. All these controllers are tuned to fulfil the same time specifications in the case of canal nominal dynamics. In some canals the plant parameters may experience large changes that deteriorate the time response, and can even unstabilize the closed-loop system. The fractional controllers are therefore designed to minimize the loss of performance of the system to variations in these parameters. Simulated results show the performance improvements achieved with these controllers compared with a conventional PI controller. © 2012 John Wiley and Sons Asia Pte Ltd and Chinese Automatic Control Society.