A construction‐integration analysis of the comprehension of causal text sequences was performed. According to the validation analysis, the comprehension of both consistent and inconsistent causal sequences requires that tentative bridging inferences be validated with reference to relevant knowledge. Consistent with this proposal, Singer (1993) reported that people needed less time to answer questions about the hypothesized validating facts after consistent and inconsistent sequences than after control, temporal sequences. The construction‐integration analysis of these effects focused particularly on a central configuration of causally related propositions. The simulations adopted the following assumptions and parameters: (a) There was a working memory buffer size of 1; (b) all link strengths had an absolute value of 1; and (c) the representations included text base and situation model elements, but no surface elements. Sixteen preliminary simulations, applied to 2 text frames each, influenced decisions concerning 4 simulation choice points. The full simulation was performed for six randomly selected texts from Singer's (1993) experiments. The results revealed a good qualitative fit between the construction‐integration activation of the probe questions and their corresponding answer times. In particular, high activation was associated with fast answer time.