O'Brien, Shank, Myers, and Rayner (1988) reported that readers generated elaborative inferences only when a text contained characteristics that virtually eliminated the possibility of an inference being disconfirmed. We reanalyzed the data of O'Brien et al. (1988) and also conducted an experiment in which we varied (a) whether or not there was an anaphoric relation between a target word and its prior mention in the text and (b) the explicitness of the prior mention. Two refinements to O'Brien et al.'s conclusions are offered. First, the two text characteristics they manipulated (a strong biasing context or a demand sentence) may have produced different types of elaborative inferencing. We argue that a biasing context results in a passive form of elaborative inferencing, involving setting up a context of interpretation, whereas the presence of a demand sentence invites the reader to actively predict a subsequent expression. Second, clear evidence for either type of inference will be apparent only with truly anaphoric materials.