The purpose of this study was to evaluate the effectiveness of two exercise modalities for improving glycosylated hemoglobin (HbA1c) and associated clinical outcomes in Polynesian adults diagnosed with type 2 diabetes and visceral obesity. Twenty-six adults were randomized to receive resistance training or aerobic training, 3×/week, for 16 weeks. Dependent variables collected before and after intervention included: diabetes markers including HbA1c, blood lipids, relevant cytokines (C-reactive protein, adiponectin), and anthropometric and hemodynamic indices. Eighteen participants (72% female; age: 49.3 ± 5.3 years; waist circumference: 128.7 ± 18.7 cm) completed the intervention and follow-up assessments. Body mass index in the whole cohort at baseline indicated Class III (morbid) obesity (43.8 ± 9.5 kg/m2). Compliance to training was 73 ± 19 and 67 ± 18% in the aerobic and resistance training groups, respectively. HbA1c remained elevated in both groups after training. Aerobic training reduced systolic and diastolic blood pressure and increased serum triglycerides (all P < 0.05). No other exercise-induced adaptations were noted within or between groups. Post hoc analysis using pooled data indicated that higher adherence to training (≥75% attendance, n = 8) significantly reduced waist circumference (P < 0.001) and tended to reduce body weight and fasting insulin (all P ≤ 0.11) versus lower adherence (<75% attendance, n = 10). In conclusion, this study did not demonstrate an improvement in HbA1c with exercise in morbidly obese Polynesian people. Future investigations involving exercise regimens that are more practicable and which involve greater frequency and duration of training may be required to induce significant and clinically meaningful adaptations in this unique diabetes population.