A limitation of the amyloid hypothesis in explaining the development of neurodegenerative diseases is that the level of amyloidogenic polypeptide in vivo is below the critical concentration required to form the aggregates observed in post-mortem brains. We discovered a novel, on-surface aggregation pathway of amyloidogenic polypeptide that eliminates this long-standing controversy. We applied atomic force microscope (AFM) to demonstrate directly that on-surface aggregation takes place at a concentration at which no aggregation in solution is observed. The experiments were performed with the full-size Aβ protein (Aβ42), a decapeptide Aβ(14-23) and α-synuclein; all three systems demonstrate a dramatic preference of the on-surface aggregation pathway compared to the aggregation in the bulk solution. Time-lapse AFM imaging, in solution, show that over time, oligomers increase in size and number and release in solution, suggesting that assembled aggregates can serve as nuclei for aggregation in bulk solution. Computational modeling performed with the all-atom MD simulations for Aβ(14-23) peptide shows that surface interactions induce conformational transitions of the monomer, which facilitate interactions with another monomer that undergoes conformational changes stabilizing the dimer assembly. Our findings suggest that interactions of amyloidogenic polypeptides with cellular surfaces play a major role in determining disease onset.