An experiment is reported in which children of two age groups (first a
nd fourth year juniors) were given a deductive reasoning task in which
the believability of conclusions was manipulated. The children were g
iven two premises with a conclusion and asked whether or not the concl
usion followed. Children of both age groups showed that their tendency
to accept the conclusions was a clear and significant function of bot
h (a) the nature of the premises, and (b) the believability of the con
clusion. They demonstrated both a positive and a negative belief bias
effect, by accepting significantly more believable than neutral conclu
sions and by accepting significantly more neutral than unbelievable co
nclusions. Older children showed significantly more premise based reas
oning and significantly less belief bias. The results are discussed wi
th reference to contemporary theories of the belief bias effect in rea
soning.