In creating educational experiences, developers often target audience segments based on demographic groups. However, we all know that people vary in other ways; one size does not fit all. This paper presents results from a research study funded by the National Science Foundation that explores the effects of three possible influences (learning style, age, and gender) on user preferences for computer-based educational activities. Using David Kolb's Experiential Learning Theory (Kolb, 1984) as a lens, we examined on-line learners' preferences for, and responses to, different types of activities ranging from deductive puzzles to open-ended design. Building on prior work presented at Museums & the Web (Schaller et al., 2002, 2005), we found that learning style does influence an individual's preferences for learning activities, particularly among adults. For example, adult social learners prefer role-play activities while intellectual learners prefer reference-style presentations. The relationship between learning styles and these preferences is stronger in adults, with adults showing more learning style-based preferences. On the other hand, among children ages 10-13 (middle school), the perceived play value of an activity has the strongest influence. While adults agree with children's play ratings, play value is not a primary consideration for adults. Age is more influential than gender in affecting activity preferences. Children prefer structured activities like Role-Play and Design. Adults prefer Interactive Reference and Puzzle-Mystery.