Gamma-cleavage of beta-amyloid precursor protein (APP) in the middle of the cell membrane generates amyloid beta protein (Abeta), and epsilon-cleavage, approximately 10 residues downstream of the gamma-cleavage site, releases the APP intracellular domain (AICD). A significant link between generation of Abeta and AICD and failure to detect AICD41-99 led us to hypothesize that epsilon-cleavage generates longer Abetas, which are then processed to Abeta40/42. Using newly developed gel systems and an N-end-specific monoclonal antibody, we have identified the longer Abetas (Abeta1-43, Abeta1-45, Abeta1-46, and Abeta1-48) within the cells and in brain tissues. The production of these longer Abetas as well as Abeta40/42 is presenilin dependent and is suppressed by {1S-benzyl-4R-[1S-carbamoyl-2-phenylethylcarbamoyl-1S-3-methylbutylcarbamoyl]-2R-hydroxy-5-phenylpentyl}carbamic acid tert-butyl ester, a transition state analog inhibitor for aspartyl protease. In contrast, N-[N-(3,5-difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester, a potent dipeptide gamma-secretase inhibitor, builds up Abeta1-43 and Abeta1-46 intracellularly, which was also confirmed by mass spectrometry. Notably, suppression of Abeta40 appeared to lead to an increase in Abeta43, which in turn brings an increase in Abeta46, in a dose-dependent manner. We therefore propose an alpha-helical model in which longer Abeta species generated by epsilon-cleavage is cleaved at every three residues in its carboxyl portion.