Purposes: This study examined the association between referral center use and outpatient medical costs among new and first-visit outpatients at a tertiary hospital, paying particular attention to whether cost differences were concentrated in diagnostic testing components. Methodology: Administrative and clinical data were analyzed for 16,447 new and first-visit adult outpatients who visited a tertiary hospital in Seoul between January 1 and January 31, 2023. The main independent variable was referral center use. Outcomes included total medical costs, general diagnostic test fees, imaging test fees, and non-test medical costs. Generalized linear models with gamma and Tweedie distributions and logarithmic links were used to estimate adjusted predicted means after controlling for age, sex, residential region, clinical department, cancer status, and Charlson Comorbidity Index category. Oaxaca-Blinder decomposition was used to assess the extent to which observable patient and clinical characteristics explained the total cost difference. Findings: Referral center users accounted for 4,553 patients, or 27.7 percent of the study population. Their unadjusted mean total medical costs were higher than those of non-users, at 175,195 versus 127,702 Korean won. After adjustment, referral center use was associated with 30.8 percent higher total medical costs but 5.2 percent lower non-test medical costs. In the adjusted predicted mean analysis, general diagnostic test fees were 42.8 percent higher and imaging test fees were 73.2 percent higher among users. Observable patient and clinical characteristics explained 6.7 percent of the total cost difference. Practical Implications: Higher total medical costs among referral center users should be interpreted alongside consideration of unmeasured clinical severity, care complexity, and tertiary-level evaluation needs. The findings suggest the need to strengthen the pre-referral sharing of test and imaging information, use prior diagnostic results, manage potentially duplicative testing, and reinforce back-referral coordination.