It has been consistently observed that people are generally overconfident in their judgments or decisions and that overconfidence in general knowledge is typically stronger among Asian than among Western subject groups. However, scientists, whose job is to produce neutral and objective knowledge, are expected to be immune to overconfidence and to be independent of crosscultural variation. The authors conducted an email survey to determine whether well-trained scholars are really immune to the overconfidence bias and whether cross-cultural variations in overconfidence exist among scientists from different language backgrounds. Contrary to the authors’ expectations, scientists who published their works in Science or Nature were inaccurate, overconfident, likely to nominate the journal they published in to have a greater impact factor,and easier to generate supportive rather than opposite arguments for their own conclusions.The authors also found that Chinese-speaking authors exhibited greater overconfidence butgenerated fewer arguments than did their English-speaking counterparts, which is consistent with Yates, Lee, and Shinotsuka’s argument recruitment model. The pervasive overconfidence and its cross-cultural variations suggest that an academic author may not communicate his or her conclusions to readers in an impartial and value-free manner, and therefore, the job of detecting bias is hoped to be taken and done well by the reader.