Research has shown general case programming to be an effective method for teaching a wide range of skills to students with disabilities. The authors used a multiple probe nested in an ABA reversal design to test (a) the effectiveness of general case programming for teaching 4 students with mild to moderate disabilities to take and give telephone messages and (b) the generality of skills across new messages and nontraining environments. The authors trained participants to take and give 1 of 4 different types of messages (student gives message to person, immediate delivery; student gives message to person, delayed delivery; student calls person, immediate delivery; student calls person, delayed delivery). The dependent variable was the percentage of task-related steps completed. Results indicate that general case programming was an efficient and effective method for teaching participants how to take phone messages. In addition, all skills generalized to new messages in nontraining environments. For 3 participants, generality data were better for taught messages compared with new messages, and 1-month follow-up data showed maintenance of skill. (Contains 1 table and 4 figures.)