The present paper addresses the basis for suppression effects observed in c
onditional inference. Previous research has shown that when an additional p
remise is included, the invalid inferences "denial of the antecedent" and "
affirmation of the consequent" are suppressed (Rumain, Connell, & Braine, 1
983). Byrne (1989) has shown that a similar suppression effect can occur in
valid inferences (modus Ponens and modus Tollens). We briefly present the
results of a series of experiments that show that strategic factors related
to task and presentation format may affect the presence of suppression eff
ects. We also present an introspective experiment (thinking aloud procedure
) that clarifies the usage of two different strategies, the integration and
the amendment strategy. The results are discussed with respect to the ment
al logic and mental model theories of reasoning.