A quasi-experimental 2-group repeated-measures design was used to test the impact of a feeding skills training program on 67 nursing assistants (treatment group n = 31; control group n = 36) regarding their knowledge, attitudes, perceived behavior control, intentions, and behaviors in feeding dementia patients. The treatment group received a feeding skills training program. Nursing assistants' knowledge, attitudes, perceived behavior control, and intentions were measured before (Pretest) and after the program (Posttest 1), and again 4 weeks later (Posttest 2). Nursing assistants (treatment group n = 20; control group n = 16) and the same number of dementia patients were measured on feeding behaviors during mealtimes before and after the training. The treatment group had significantly more knowledge (P < .001), greater intention to feed (P = .05), and better behaviors toward feeding dementia patients (P = .009) than the control group. There were no significant differences between the groups in attitude (P = .85), intention beliefs (P = .11) or perceived behavior control (P = .99). Thus, the intervention was effective at changing knowledge, intention, and behaviors among nursing assistants.