Background: Poor oral health at hospital admission is a potential higher mortality risk predictor. We aimed to determine in-hospital mortality by assessing poor oral health using a validated tool. Methods: A retrospective observational study was conducted in an acute care hospital, and 624 consecutive geriatric patients were included. Patients were divided into three groups according to oral health, stratified by the Oral Health Assessment Tool (OHAT) scores. Nutritional status, daily living activities, cognitive impairment, and comorbidities were collected as covariates. Univariate and multivariate analyses were performed to identify the relationship between oral health and survival. Results: The mean age was 83.8 ± 7.9 years, and 41% were males. Groups with an OHAT score equivalent to 0, 1-2, and ≥ 3 comprised 213, 206, and 205 patients, and 11 (5.2%), 13 (6.3%), and 37 (18.0%) of those patients died in the hospital, respectively. Patients in the OHAT score ≥ 3 group had higher mortality than those in the other groups (log-rank test: p = 0.012 for the OHAT = 0 group; p = 0.010 for the OHAT = 1-2 group after Bonferroni corrections). Patients in the OHAT score ≥ 3 group continued to have poor survival even after adjusting for confounders in the Cox's regression analysis (hazard ratio: 2.514, 95% confidence interval: 1.220-5.183, p = 0.012). Conclusion: In geriatric patients, poor oral health at hospital admission was an independent in-hospital mortality predictor. Future studies on oral care intervention stratified by oral health conditions are warranted.