Amyloid-β (Aβ) is believed to directly affect memory and learning in Alzheimer's disease (AD). It is widely suggested that there is a relationship between Aβ40 and Aβ42 levels and cognitive performance. In order to explore the validity of this relationship, we performed a meta-analysis of 40 peer-reviewed, published AD transgenic mouse studies that quantitatively measured Aβ levels in brain tissue after assessing cognitive performance. We examined the relationship between Aβ levels (Aβ40, Aβ42, or the ratio of Aβ42 to Aβ40) and cognitive function as measured by escape latency times in the Morris water maze or exploratory preference percentage in the novel object recognition test. Our systematic review examined five mouse models (Tg2576, APP, PS1, 3xTg, APP(OSK)-Tg), gender, and age. The overall result revealed no statistically significant correlation between quantified Aβ levels and experimental measures of cognitive function. However, enough of the trends were of the same sign to suggest that there probably is a very weak qualitative trend visible only across many orders of magnitude. In summary, the results of the systematic review revealed that mice bred to show elevated levels of Aβ do not perform significantly worse in cognitive tests than mice that do not have elevated Aβ levels. Our results suggest two lines of inquiry: 1) Aβ is a biochemical “side effect” of the AD pathology; or 2) learning and memory deficits in AD are tied to the presence of qualitatively “high” levels of Aβ but are not quantitatively sensitive to the levels themselves.