Dietary habits play an important role in the development of obesity and type 2 diabetes. However, evidence on association between diet frequency and type 2 diabetes was limited and inconclusive. We aimed to examine the association between meal frequency and risk of type 2 diabetes. The cohort study used data from China Health and Retirement Longitudinal Study of 8874 community-dwelling people aged over 45 years. Participants were classified as eating 2 meals/d, 3 meals/d and 4 meals/d. Multiple Poisson regression models were used to examine risk of 4-year incident type 2 diabetes among people who ate more or less than 3 meals/d compared with people who ate 3 meals/d. We documented 706 type 2 diabetes cases during follow-up. After adjustment for known risk factors for type 2 diabetes, except for body mass index (BMI), participants who ate 4 meals/d were at a lower risk of type 2 diabetes than those who ate 3 meals/d (RR = 0.73 [0.58, 0.92]). After further adjustment for baseline BMI, the association was slightly attenuated but remained statistically significant (RR = 0.76 [0.60, 0.97]). Subgroup analysis showed that the fully-adjusted RRs of type 2 diabetes for people eating 4 meals/day were 0.66 (0.48, 0.91) and 0.93 (0.65, 1.34) among those had a BMI < 25 and ≥ 25 kg/m2, respectively. Eating 4 meals/day, compared with eating 3 meals/day, was associated with lower risk of developing type 2 diabetes in a Chinese population, particularly in those with a BMI < 25 kg/m2.