Dynamic instructional planning was explored as a control mechanism for intelligent tutoring systems through the development of the Blackboard Instructional Planner--a blackboard software-based dynamic planner for computerized intelligent tutoring systems. The planner, designed to be generic to tutors teaching troubleshooting for complex physical devices, generates a sequence of lesson plans customized to a student model inferred from a pre-instruction questionnaire. Content, delivery, and length of lessons are determined by the inferred student model and revised during instruction in response to student questions and requests, changes in time remaining for lessons, and modifications to the student model. The planner has been implemented as the controller for the Lower Hoist Tutor, a prototype tutor for the Mark-45 naval gun mount, which teaches troubleshooting in the hoist assembly. The more economic representation of pedagogical knowledge in the planner facilitates extension to new domains compared to traditional computer assisted instruction systems which procedurally encode pedagogical decisions. Twenty-eight figures illustrate the text. Five appendixes give additional information on system development and performance. (SLD)