The key to successful control of diabetes mellitus is owed largely but not limited to self-care practices by patients. The purpose of this study was to identify factors that are associated with the increase in hospital admissions among diabetes mellitus patients in Guruve and Mazowe districts. An analytic cross-sectional study was conducted at Guruve, Mvurwi, Howard, and Concession hospitals. A census of all the 202 registered patients attending diabetes mellitus review appointments at treatment centers was done. Interviewer-administered pretested questionnaires were used to collect data. The admission prevalence was 43.6%. Insulin medication [AOR 2.74 (95% CI: 1.22;7.27) p = 0.0168], low educational level (primary and below) [AOR 2.74 (95% CI: 1.34; 5.58) p = 0.0057], having long been diagnosed with diabetes(1 year+) [AOR 3.06 (95% CI: 1.38; 6.79) p = 0.0057], and residing in areas where there is no diabetes mellitus education at community level [AOR 3.86(95% CI: (1.60;9.32) p = 0.0026] were independent factors associated with hospital admissions due to diabetes mellitus in Guruve and Mazowe districts that remained statistically significant after logistic regression analysis. Independent predictors of hospitalization were: type of treatment, educational level, accessing community-based health education, and time duration of diabetes diagnosis. Access to services and insulin supplies for diabetic patients was not adequate. Health managers at district hospitals will need to improve access to such services for diabetic patients. Insulin medication was associated with readmissions. There is a need to improve self-care practice using an ecological approach that involves the family and community.