Providing students an appropriate level of challenge, called the instructional level, is an important component of effective instruction. Research regarding the optimal ratio of known to unknown items for drill tasks has been inconsistent. The author of the current article conducted an empirical meta-analysis of research on drill ratios by using the following groupings: < 50% known, 50% to 69% known, 70% to 85% known, and 90% known. Fifty-five effect size coefficients were computed, after eliminating outlying data, from 13 articles, with the 3 least challenging ratios resulting in strong mean effects. The most challenging (< 50% known) ratio resulted in a small to moderate effect. A mean effect size of .82 computed for various student outcomes compared to .43 for student preferences. Much stronger effects were also noted for tasks involving acquisition of new skills compared to proficiency tasks. Implications for research and practice are discussed.