ABSTRACT: Although immediate early genes (IEGs) such as Bdnf, Arc and Egr1, have been implicated in plasticity, the larger pathways related to memory and memory disorders are not well understood. Here, we combined statistical Affymetrix microarray and behavioral analyses to identify key genes and pathways associated with aging-related cognitive impairment. Aged rats were separated into cognitively unimpaired (AU) or impaired (AI) groups, based on their Morris water maze performance relative to young-adult (Y) animals. Hippocampal gene expression was assessed in Y, AU and AI on the fifth (last) day of maze training or 21 days posttraining, and in non-trained aged and young animals (eight groups, overall n = 78, one chip/animal). ANOVA, linear contrasts, and overrepresentation analyses identified genes and pathways that differed from Y generally with aging (in both AU and AI) or selectively with cognitive status (only in AI or AU). Plasticity pathways, including insulin/cAMP/IEG signaling, and glycogenolytic and lipogenic pathways, were selectively downregulated (5 days) in AI, whereas Notch2 (regulating oligodendrocyte differentiation) and myelination pathways were upregulated (particularly at 21 days). Downregulation with general aging occurred in signal transduction and axonal growth/transport pathways, whereas upegulation occurred in immune/inflammatory, lipid metabolism/transport (e.g., Lxr-Srebf1), and lysosomal pathways. In AU, receptor/signal transduction genes were selectively upregulated, suggesting possible compensatory mechanisms. Immunohistochemistry confirmed and extended results to the protein level. Thus, this study identified novel cognition-linked processes, suggesting a new model in which energy-intensive, plasticity/lipogenic processes and energy-generating pathways necessary for learning are coordinately downregulated during training, while myelinogenic programs that impair cognition are concurrently activated. Experiment Overall Design: Aged rats were separated into cognitively unimpaired (AU) or impaired (AI) groups, based on their Morris water maze performance relative to young-adult (Y) animals (NT, 5D, and 21D, N=10/group). Hippocampal gene expression was assessed in Y, AU and AI on the fifth (last) day of maze training or 21 days posttraining, and in non-trained aged and young animals (eight groups, overall n = 78, one chip/animal)