The present independent scientific study was designed to evaluate the electromyographic (EMG) activity during abdominal exercise performed using a commercially available device, the AbRoller, as compared with that performed conventionally. The relative peak and mean EMG activities of the rectus abdominis, rectus femoris, sacrospinalis, and sternocleidomastoid were measured during the two modes of abdominal exercises using surface electromyography. Forty subjects (21 females and 19 males) aged 20.8 ± 3.6 years (body mass index = 27.1 ± 4.8 kg/m2; waist circumference = 0.8 ± 0.1 m) performed 5 repetitions of 45-degree trunk curls and 5 repetitions of 90-degree trunk curls on both the AbRoller and the conventional abdominal exercises. To compare the differences in the relative peak and mean EMG activities between the two exercise modes, repeated measures two-way ANOVA (mode x angle) were utilized. For the sternocleidomastoid muscle, the AbRoller-performed abdominal exercises resulted in significantly (p < .05) lower relative peak and mean EMG activities than those of the conventional abdominal exercises. For the rectus abdominis, rectus femoris, and sacrospinalis muscles, no significant (p > .05) differences in the relative peak and mean EMG activities were found between the two exercise modes. The 90-degree trunk curls generated significantly (p < .05) higher relative peak and mean EMG activities in rectus femoris than the 45-degree curls in both exercise modes. In conclusion, the AbRoller and the conventional abdominal exercises are similarly effective in recruiting the abdominal muscle whereas the AbRoller exercise offers better neck support. Hip flexor and back muscular involvements are similar between the AbRoller and the conventional abdominal exercises. Regardless of exercise modes, the angular displacement of the trunk curl should be limited within the initial 45 degrees to minimize the undesirable hip flexor activation. Keyword(s): exercise/fitness, physical activity