An experimental research program was carried out in order to validate a methodology for the mechanical assessment of structures damaged by alkali-silica reaction (ASR). Long-term deformations of specimens kept under varied environments were measured. Results were collected according to the French methodology for measuring residual expansion, on cores drilled from cylinders and prisms made of the same concrete materials. Numerous testing conditions were intended to check the experimental robustness of the most significant input data for structural assessment. It appears that omitting the measured expansion of the specimens during a rather long period (60 days), as presently recommended, can lead to an unsafe estimation of ASR-residual expansions, especially when the core is drilled during the active phase of ASR development. The strains of the specimens during the first days of the test are caused by the superimposition of concrete expansion due to water sorption and additional ASR expansions. Evolutions and improved conditions of application relative to the French standard method for estimating residual expansion of ASR-affected concrete are proposed.