It is important for the teacher to choose effective software for students. It is also important for designers to know what features of educational software make it effective. But the results of studies dealing with the effectiveness of educational software are contradictory. One reason for such results might be the fact that meta-analysis covers different kinds of educational software, simulations and drills for example, but it is hard to compare them. Another reason might be that the conditions (instruments, subjects, etc.) of the various studies were different. A third possibility might be that a particular type of educational software could consist of both good and bad characteristics. Therefore, research is needed on the characteristics of educational software. This type of study focuses on a particular kind of educational software, and this article is about the findings of one such study, exploring the characteristics and the effectiveness of drills in Estonian elementary schools. Sixteen characteristics of drills that had statistically significant correlation with student learning, and the influence of these characteristics are discussed as the results of this study. The results indicated that characteristics related to feedback, presentation of information and questions are the most critical evaluating quality of drills and in designing such educational software.