This work is part of an overall project for the reassessment of concrete structures damaged by alkali-silica reaction (ASR). The paper focuses on developing a laboratory method for expansion tests since the usual tests appear to be difficult to use in expert assessment. The development involves optimising the storage conditions and the sizes of the specimen and aggregate. A combined effect of the aggregate and specimen sizes on ASR expansion is thus pointed out: for a given mortar, the expansion is lower in small specimens than in large specimens. Therefore, the ratio `specimen size/aggregate size' has to be sufficiently high to decrease this scale effect and obtain relevant measurements. The discussion proposes a method for always using comparable conditions during the expansion tests and finally suggests how this test can be optimised to provide fast and relevant results for use in structure reassessment.