The univariate survival analysis revealed a 5-year RFS of 72.5%, with a median 5-year RFS of 8.4 years. The diabetic patients showed a reduced 5-year RFS in contrast to the non-diabetic cases (64.0% vs. 74.5%, p = 0.016; Figure 2). In contrast, a routine metformin treatment did not alter the RFS in the complete cohort of HNM patients (p = 0.805; Figure 1). Besides the univariate survival analysis, we performed a multivariate Cox regression to adjust for covariables such as age, BMI, statin use, intake of ASA, tumor size, nodal status, ulceration, and distinct tumor histologic types. For the entire cohort of HNM patients, diabetic cases showed, in line with the results of the univariate statistics, reduced RFS (HR = 1.980, 95% CI = 1.108–3.538, p = 0.021; Table 3). For metformin use, the multivariate Cox regression analysis resulted in a significantly more favorable outcome in comparison to the controls (HR = 0.396, 95% CI = 0.177–0.884, p = 0.024; Table 3). Additionally, we analyzed distinct recurrence rates in detail, distinguishing a cumulative recurrence rate, a cumulative locoregional recurrence rate, as well as a cumulative distant recurrence rate (Table 3). Especially when focusing on the cumulative locoregional recurrence rate, both effects, the detrimental one of DM2 (HR = 4.173, 95% CI = 1.628–10.697, p = 0.003) as well as the beneficial one of a routine metformin use (HR = 0.135, 95% CI = 0.031–0.589, p = 0.008) were even more pronounced (Table 3). Besides the survival analysis of the complete cohort of HNM patients, we performed a more detailed perspective analysis on the cohort of diabetic patients. Hereby, univariate Cox regression displayed a significantly beneficial effect of metformin use on the RFS of diabetic HNM patients (HR = 0.417, 95% CI = 0.201–0.868, p = 0.019; Table 4). After applying multivariate statistics, this beneficial effect on the RFS of HNM patients was validated in the subgroup analysis (HR = 0.352, 95% CI = 0.135–0.913, p = 0.032; Table 4).