Abstract A believable conclusion is usually judged more acceptable than an unbelievable one, all other things being equal. However, there has been little empirical work to address how the believability of the premises affects the acceptability of an argument. In the present experiment, participants solved problems having either believable, unbelievable, or neutral premises, and having either believable or unbelievable conclusions. People were more likely to accept a conclusion when it was supported by believable premises than when it was supported by either unbelievable or neutral premises; this effect was true of both valid and invalid arguments. The fact that premise believability did not interact with logical validity suggested that premise believability acts independently of logical analysis. The results suggest a filtering mechanism, which operates after logical analysis has occurred, and which rejects conclusions that are unbelievable, or that are derived from unbelievable premises. By definition, a sound argument is one that has good form (i.e., is logically valid) and good content (i.e., is based on true premises). A sound argument produces conclusions that are always true, and which follow necessarily from their premises (Johnson & Blair, 1983; Kelley, 1994); as such, the ability to construct sound arguments is one of the hallmarks of rational thought. It is surprising, therefore, that a systematic investigation of how people evaluate the soundness of arguments has not been carried out. Specifically, whereas there is a well - developed literature concerning how the validity of arguments is evaluated (see Braine & O'Brien, 1991; Johnson - Laird, Byrne, & Schaeken, 1992 for recent reviews), very few studies have investigated how the of the premises affects reasoning performance. Consequently, the purpose of this article was to test the hypothesis that the acceptability of a deductive argument varies as a function of the of its premises. For the purposes of this article, truth will be operationally defined in terms of believability: that is, a premise will be deemed to be true to the extent that the reasoner believes it to be true. The few studies that have investigated the believability of the premises have yielded contradictory findings. For example, Revlis (1974) found that the believability of the premises did not alter the manner in which people made logical decisions, and thus concluded that this factor played a minor role in people's reasoning judgments. In contrast, at least two investigations (e.g., Hawkins, Pea, Glick, & Scribner, 1984; Markovits & Vachon, 1989) have found that both children and adolescents were more likely to make logically correct inferences when they were provided with believable as opposed to unbelievable premises. However, neither set of studies provides conclusive data. Revlis did not manipulate believability directly, and the materials he used had a restricted range of believability, reducing the probability of observing effects of this variable. Similarly, in the latter two studies, the believability of the premises was not manipulated independently of the believability of the conclusions, so that one cannot draw firm conclusions regarding the effects of premise believability per se. Although studies on the effects of premise believability have not produced clear results, related findings suggest that this factor is a plausible source of variability in deductive reasoning. First, other premise - based interpretational factors play an important role in how people evaluate arguments. Among other things, inference patterns vary as a function of how the logical quantifiers are interpreted (Begg & Harris, 1982), whether the premises are converted or not (i.e., whether statements such as All A's are are interpreted to mean All B's are A's; Newstead, 1989), and how necessity/sufficiency relations are interpreted (Thompson, 1994; 1995). …