The implication of soluble Abeta in the Alzheimer's disease (AD) pathology is currently accepted. In fact, the content of soluble extracellular Abeta species, such as monomeric and/or oligomeric Abeta, seems to correlate with the clinico-pathological dysfunction observed in AD patients. However, the nature (monomeric, dimeric or other oligomers), the relative abundance, and the origin (extra-/intraneuronal or plaque-associated), of these soluble species are actually under debate. In this work we have characterized the soluble (defined as soluble in Tris-buffered saline after ultracentrifugation) Abeta, obtained from hippocampal samples of Braak II, Braak III-IV and Braak V-VI patients. Although the content of both Abeta40 and Abeta42 peptides displayed significant increase with pathology progression, our results demonstrated the presence of low, pg/µg protein, amount of both peptides. This low content could explain the absence (or below detection limits) of soluble Abeta peptides detected by western blots or by immunoprecipitation-western blot analysis. These data were in clear contrast to those published recently by different groups. Aiming to explain the reasons that determine these substantial differences, we also investigated whether the initial homogenization could mobilize Abeta from plaques, using 12-month-old PS1xAPP cortical samples. Our data demonstrated that manual homogenization (using Dounce) preserved the integrity of Abeta plaques whereas strong homogenization procedures (such as sonication) produced a vast redistribution of the Abeta species in all soluble and insoluble fractions. This artifact could explain the dissimilar and somehow controversial data between different groups analyzing human AD samples.