Different aspects of text comprehension and different evaluation procedures were studied for elementary school students. Indicators sensitive to the different levels of comprehension involved were used, and recall was divided into recall macro, micro, and total. The sample consisted of 90 sixth graders in the national capital region of Canada. Relationships among the results of a free, written recall protocol, a questionnaire, and problem-solving activities were examined as they allowed the observation of different text processing levels. The problem-solving activities measured something different than the recall task and questionnaire measures; no significant correlations were obtained between these activities and other variables in the study. The questionnaire did not measure exactly what the recall task measured, and the recall task appeared more sensitive to the effects of test type and domain knowledge. Overall, the recall task was a more discerning performance measure. The recall task and the problem-solving activities appear to be the most promising approaches for constructing a reading comprehension text for sixth-grade native speakers of French. Three bar graphs and two tables present study data. (Contains 19 references.) (SLD)