Alzheimer’s disease (AD) is the most common type of dementia that affects the lives of nearly 50 million people globally. No efficient treatment or cure has been found for AD yet. Different studies have been performed on heterogeneity in AD from various perspectives. Analyzing AD as a disease with several subtypes might assist us to identify effective personalized treatments for AD patients. Here, we investigated the heterogeneity in AD from a molecular perspective. In particular, we used gene expression profiles of 93 samples from the hippocampus of AD patients. To increase the sample size and enhance the robustness of the results, we applied meta-analysis on three different datasets. We first combined gene expression profiles from multiple relevant AD studies, and then, using a clustering analysis, we found two distinct molecular subtypes across AD samples. We validated the obtained results using various statistical and randomization approaches. Furthermore, we showed that the two transcriptomic subtypes are statistically independent of sex, age and Braak stage of the subjects. Notably, one of the identified subtypes indicates high similarity to control samples, which suggests that AD onset does not necessarily affect the hippocampal transcriptome. By performing differential gene expression and pathway enrichment analyses for various settings, we found 2371 differentially expressed genes that can be used for discriminating between the two AD subtypes. Finally, using a clustering approach, we introduced a novel molecular signature for the two subtypes based on14 differentially expressed genes.