Type 2 diabetes is an emerging concern in Kenya. This clustered-randomized trial of peri-urban communities included a theory-based and culturally sensitive intervention to improve diabetes knowledge, health beliefs, dietary intake, physical activity, and weight status among Kenyan adults. Those in the intervention group (IG) received a culturally sensitive diabetes education intervention which applied the Health Belief Model in changing knowledge, health beliefs and behavior. Participants attended daily education sessions for 5 days, each lasting 3 h and received mobile phone messages for an additional 4 weeks. The control group (CG) received standard education on COVID-19. Data was collected at baseline, post-intervention (1 week), and follow-up assessment (5 weeks). Linear mixed effect analysis was performed to assess within and across group differences. Compared to the control, IG significantly increased diabetes knowledge (p < 0.001), health beliefs including perceived susceptibility (p = 0.05), perceived benefits (p = 0.04) and self-efficacy (p = 0.02). IG decreased consumption of oils (p = 0.03), refined grains (p = 0.01), and increased intake of fruits (p = 0.01). Perceived barriers, physical activity, and weight status were not significantly different between both groups. The findings demonstrate the potential of diabetes education in improving diabetes knowledge, health beliefs, and in changing dietary intake of among adults in Kenya.