Reading evaluation should (1) be a continuous process using many samples of a learner's work, (2) allow students to engage in uninterrupted reading, (3) constitute a learning situation, and (4) indicate new directions for improvement in an open-ended manner. Current practices for evaluating children's reading processes are usually based on samples of work that are part of the basal reading series and offer only a limited view of students' abilities. Basal series also stress a mastery learning approach, which leaves little time for the uninterrupted reading needed for a fair evaluation of reading processes. Finally, students who fail to meet the criteria to progress to the next level in a basal reading program are offered more basal reading rather than new avenues for improvement. A more accurate alternative, naturalistic assessment is based on observing students' responses to reading as they are recorded through devices such as checklists, anecdotal records, tape recording, and work samples. These procedures can be structured as learning activities in the course of teacher-student interviews. Students can respond individually or interact with a group in response to a text, and by carefully structuring these interviews, teachers can make inferences about students' covert thinking processes in reading. This and other adaptations of the interviewing process fulfill the criteria for effective reading evaluation. (Categories of reading strategies are included.) (SRT)