Use of an alpha-lattice design in 714 international yield trials of barley, durum wheat and bread wheat in 1990/91 to 1992/93 resulted in an average efficiency 18% higher than the randomized complete block design when average variance was used as the comparison criterion. Alpha-lattice was generally most efficient when the C.V.s of the trials were high. It was also slightly more efficient for low-yielding than for high-yielding trials, and for rainfed than for irrigated trials. Since the changeover to alpha-lattice designs requires no new major input or changes in present field layout, its use for international trials is recommended.