The effectiveness of conducting computer-based training (CBT) at Coast Guard duty stations was determined in an experiment, comparing 3 training delivery approaches: 1) nonresident CBT course conducted at students' duty stations; 2) resident CBT course (the same course) conducted at a training center; and 3) traditional resident instructor-led course conducted at a training center. The existing 1-week AN/WSC-3 UHF transceiver maintenance course was selected for the evaluation, with the CBT version developed by an independent contractor. Training effectiveness was determined using several measures, including student hands-on performance during the troubleshooting and repair of actual malfunctioning transceivers, pre- and post-training knowledge tests, and student reactions to the training. The nonresident CBT course conducted at duty stations was found to be as effective as the instructor-led resident course, and required substantially less training time. Follow-up interviews conducted with a subset of the participating duty stations found that students, commanders and other staff strongly support the implementation of nonresident training. They also found that staff were concerned about how nonresident CBT would be implemented; important issues and potential obstacles were identified. A cost analysis uncovered trade-offs between resident instructor-led and nonresident CBT courses, and found that nonresident CBT provided to students at duty stations can save training dollars.