We review studies testing performance on tasks putatively tapping spatial memory in the Tg2576 mouse model of Alzheimer's disease (AD). This model exhibits age-dependent elevation of amyloid-β in the hippocampal formation and elsewhere in the brain. From 49 articles in all, we reviewed Tg2576 performance on five spatial memory tasks: the reference memory version of the Morris water maze, continuous Y-maze alternation, discrete forced-choice T-maze alternation, the radial arm water maze, and the circular platform maze (Barnes maze). Proportionally, the likelihood of detecting significant impairment in Tg2576 mice (relative to age-matched controls) was found to be: highest with the use of T-maze alternation and the radial arm water maze; intermediate when using the Morris water maze and continuous Y-maze alternation; and lowest when using the circular platform maze. These results are indicative rather than conclusive, but have implications for testing cognitive function in Tg2576 mice and, potentially, other AD rodent models. The apparent sensitivity of the T-maze alternation task and reduced sensitivity of the Morris water maze task (reference memory version) are discussed. We also consider limitations and potential improvements in assessing cognitive impairment in dementia models.