To weigh the negative effects of "primerese" against its supposedly improved readability, a study compared the reading of controlled basal stories with that of the same stories minimally rewritten into more natural language. Participants were 64 white, lower middle class first grade students selected from five classrooms (all with basal instruction programs) in two communities. The children were paired by word recognition scores, each member reading alternate versions of the same four stories over a two-week period. They were tested for comprehension and for oral reading miscues. In comprehension, subjects with high word recognition scores profited from changes in the revised versions, while low word recognition subjects were not affected. Students reading the rewritten version answered a majority of questions correctly more often, suggesting that reduced dependence on pictures and natural language conventions facilitated inference, coordination of different parts of the text, and a greater use of expectations based on natural language in top-down processing of the texts. Analysis of oral reading miscues indicated that while the rewritten versions had no significant effect on rate of reading or word miscues, they reduced the proportion of graphically based miscues for high word recognition subjects, who consistently produced fewer punctuation miscues, suggesting that the rewritten texts were more in keeping with expectations children derive from naturally occurring language patterns. Tables, references, texts and test questions are included. (JG)