A one year descriptive study investigated the strategies used by a group of competent beginning readers who were learning to read in a program in which text reading was the main instructional task. Subjects included 52 six-year-old children attending 20 randomly selected schools in Auckland, New Zealand. Strategy usage in oral reading was related to three factors: learning opportunities over time, reading proficiency, and text difficulty. The data, comprised of errors and self-corrections, were analyzed at three linguistic levels--graphemic, syntactic, and semantic--to infer strategies. The progress the children made on code-emphasis texts indicated their improvement in decoding skills. Analysis of the errors showed that the children made wide use of syntactic and semantic cues to help them learn an efficient system of responses to graphemic cues. Evidence of self-correction strategies was also found. The results of the study highlight the complex nature of learning to read and suggest that while some teaching programs that emphasize a particular learning strategy may be successful, they may not take full advantage of other equally valid strategies that children can develop. (HOD)