Transforminganalog graphic data, such as maps, into digital format by manual digitizing is slow and expensive, but is nevertheless widely performed.Studies of digitizing methods to find opportunities for optimization are therefore warranted.The work reported here investigates the possibility of applying the Keystroke-Level Model to the modeling and optimization of manual map digitizing tasks.We tested the suitability of the model for manual digitizing at a national mapping agency and determined unit tasks with their performance times.The paper describes the design of an experiment to measure performance times under production conditions.Two new keystroke level operators are defined for manual digitizing.The use and suitability of the model are demonstrated by analyzing the differences between predicted and measured performance times for unit tasks.The results confirm the applicability and the economic importance of keystroke-level analyses of real world tasks.