This study examines the effect of item response time across 30 items on ability estimates in a high stakes computer adaptive graduate admissions examination. Examinees were categorized according to 4 item response time patterns, and the categories are compared in terms of ability estimates. Significant differences between response time patterns were observed. Highest ability estimates were associated with a response time pattern that was consistent across items, while lowest ability estimates were associated with long response times on items early in the test and short response times late in the test. These results suggest that teaching examinees to manage time effectively can maximize ability estimates. An alternative interpretation is that more able examinees require less time to respond to items. (Contains 4 tables and 1 figure.)