A range of genetic mouse models of Huntington's disease have been created. However, as knock-in models typically have milder phenotypes, they have frequently been overlooked as therapeutic tools in favour of the transgenic models that display severe behavioural symptoms. More sensitive tests are therefore required to reveal abnormalities and release the potential of knock-in lines. An implicit learning task for mice has been developed in the nine-hole operant box test apparatus, in which mice must respond to a sequence of lights in order to earn a reward. A light stimulus was presented randomly in one of five holes, to which a nose poke response resulted in the light being extinguished and a second light illuminated in a different hole. Response to the second light resulted in a liquid reward. To probe implicit learning, a predictable stimulus sequence was embedded among many unpredictable sequences. In the current study, the Hdh(Q92) mouse model of Huntington's disease was examined. At 4 months of age, Hdh(Q92/Q92) mice demonstrated clear and significant deficits in both accuracy and reaction time on all trials of the implicit learning task, with improved performance on predictable trials. We believe this to be the earliest reported behavioural deficit in the Hdh(Q92) knock-in mouse line. The results of this study validate the serial implicit learning task as a sensitive tool for the examination of implicit and motor learning deficits in mutant mice, and may provide a powerful test for probing potential treatments for Huntington's disease.