Background: Diabetes health belief scale was developed to measure attitudes about diabetes self care which could help to explain compliance with prescribed medical regimens. The degree to which patients follow advice as self-care behaviours is determined by health belief of diabetes. Objective: To assess diabetes health beliefs of type 2 adult diabetic patients attending chronic follow up units of Dessie referral hospital. Method: Cross-sectional study design was conducted from May 1 to 30, 2018 in Dessie referral hospital diabetic chronic follow up units. Systematic random sampling was a method to select a total of 278 type 2 diabetic patients by using patient registration book as a sampling frame. Data was collected by interview. Epidata 3.1 and SPSS version 23 software were applied for data entry and analysis respectively. Tables and graphs were used to present descriptive analysis. A bivariate logistic regression analysis was done to see the crude association between independent and outcome variables. After that multiple logistic regression model was done by selecting only variables with P-value <0.2 in bivariate analysis. Significant statically association were considered based on adjusted odd ratio included in 95% confidence interval at P-value less than 0.05. Results: Among respondents 43.5% had low aggregate diabetes health belief and the rest 56.5% had high diabetes health belief. From those variables which were entered to multivariate analysis model marital status, attended a diabetic education, perceived susceptibility, perceived severity, perceived benefit and perceived barrier were found to be significantly statistical associated with aggregate diabetes health belief. Conclusion and recommendations: Diabetes health beliefs of patients were not adequate so that there is a need to improve diabetes health beliefs of patients. Attention should be given towards diabetes health belief by Dessie referral hospital administrators, health care professionals, diabetic associations and researchers.