An important component of the pathologic process underlying Alzheimer's disease is oxidative stress. Met(35) in amyloid beta-protein (A beta) is prone to participating in redox reactions promoting oxidative stress, and therefore is believed to contribute significantly A beta-induced toxicity. Thus, substitution of Met(35) by residues that do not participate in redox chemistry would be expected to decrease A beta toxicity. Indeed, substitution of Met(35) by norleucine (Nle) was reported to reduce A beta toxicity. Surprisingly, however, substitution of Met(35) by Val was reported to increase toxicity. A beta toxicity is known to be strongly related to its self-assembly. However, neither substitution is predicted to affect A beta assembly substantially. Thus, the effect of these substitutions on toxicity is difficult to explain. We revisited this issue and compared A beta 40 and A beta 42 with analogs containing Met(35)-->Nle or Met(35)-->Val substitutions using multiple biophysical and toxicity assays. We found that substitution of Met(35) by Nle or Val had moderate effects on A beta assembly. Surprisingly, despite these effects, neither substitution changed A beta neurotoxicity significantly in three different assays. These results suggest that the presence of Met(35) in A beta is not important for A beta toxicity, challenging to the prevailing paradigm, which suggests that redox reactions involving Met(35) contribute substantially to A beta-induced toxicity.