Transgenic animals expressing mutant human amyloid precursor protein (APP) are used as models for Alzheimer disease (AD). Ideally, behavioral tests improve the predictive validity of studies on animals by mirroring the functional impact of AD-like neuropathology. Learning and memory studies in APP transgenic models have been difficult to replicate. Standardization of procedures, automatization or improved protocol design can improve reproducibility. Here the IntelliCage, an automated system, was used for behavioral testing of APP female transgenic mice with both the Arctic and Swedish mutations, the tg-ArcSwe model. Protocols covering exploration, operant learning, place learning and extinction of place preference as well as passive avoidance tests were used for longitudinal characterization of behavior. Differences in exploratory activity were significant at four months of age, when plaque-free tg-ArcSwe mice visited less frequently the IntelliCage corners and initially performed fewer visits with licks compared to non-tg animals, inside the new environment. Fourteen months old tg-ArcSwe mice required a longer time to re-habituate to the IntelliCages than non-tg mice. At both ages tg-ArcSwe mice perseverated in place preference extinction test. Fourteen months old tg-ArcSwe mice were impaired in hippocampus-dependent spatial passive avoidance learning. This deficit was found to inversely correlate to calbindin-D28k immunoreactivity in the polymorphic layer of the dentate gyrus. Reduced water intake and body weight were observed in 4 months old tg-ArcSwe animals. The body weight difference increased with age. Thus behavioral and metabolic changes in the tg-ArcSwe APP model were detected using the IntelliCage, a system which provides the opportunity for standardized automated longitudinal behavioral phenotyping.