Forty-three college students read a specially prepared text either with or without inserted questions. The text and the questions were presented on a computer terminal to allow measurement of reading times on short segments of material. Question groups performed better, relative to controls, on posttest items that repeated inserted questions and also on new posttest items from the same categories as the inserted questions. While there was no overall reading time difference between the question and no-question groups, subjects who received inserted questions spent more time on the parts of the text that contained information of the type needed to answer the questions. The results are consistent with a selective attention interpretation of the indirect effect of inserted questions. (Author)