The current method of diagnosing diabetes mellitus (DM) using fasting plasma glucose (FPG) and hemoglobin A1c (HbA1c) without oral glucose tolerance tests (OGTT) results in inaccuracy and underdiagnosis of diabetes. This study aimed to investigate the mortality of individuals with improperly classified glycemic status and suggest new screening methods for diabetes. A total of 1935 subjects were prospectively followed for 10 years, with vital status obtained through linkage to the National Health Insurance Research Database (NHIRD). Hazard ratios (HR) of all-cause mortality were analyzed. The percentage of all-cause mortality was significantly higher in subjects with improperly classified glycemic status than those correctly classified (8.3% vs. 3.6%, p = 0.004; unadjusted HR 2.537, p = 0.002). The risk of mortality associated with underdiagnosis was highest in those over the age of 60 (adjusted HR 2.043, p = 0.036). Using components of metabolic syndrome, three screening strategies were developed to determine the need for OGTT in elderly subjects with improved sensitivity in diagnosing DM. Diagnosis of diabetes based solely on FPG and HbA1c leads to underdiagnosis of glycemic status and higher all-cause mortality in older adults. Age and the presence of components of metabolic syndrome-based screening strategies can improve diagnostic accuracy and reduce the need for OGTT.