This study tested the feasibility of computer-based training (CBT) for students with below average academic skills, and evaluated a job performance aid used in recipe conversation for the Mess Management Specialist School. There were 20 students in each of three groups: two CBT experimental groups and a control group. One CBT group received the job aid and the other, the traditional math as taught in the school but on-line. The control received classroom training from an instructor using traditional math. The CBT groups required significantly less training time than the control group while maintaining high scores on the final test which was criterion referenced. However, the CBT job aid group performed significantly poorer than the no job aid and classroom groups. The student questionnaire indicated that the typical student gave a high rating to the CBT instruction. It is suggested that individualized CBT is efficient in terms of reduced training time. (Author)