An experiment was designed to test the interactive reading theory prediction that children's reading strategies vary with the availability of higher-level information. Third- and fourthgrade children (aged 8 and 9 years, respectively) were assigned to one of three context conditions and one of three experimental instructions conditions. Results supported the hypothesis the children adopt different oral reading speed strategies, depending on the amount of contextual information available, in order to maximize reading accuracy and, where appropriate, ongoing comprehension. When reading meaningful materials, children adopt a fluent, top-down reading strategy that is relatively resistant to modification. Although a slower, bottom-up strategy is preferred in the reading of words in isolation, children are able to modify this strategy in accordance with experimental instructions. These results are consistent with an interactive theory of reading.