Swallowing dysfunction caused by stroke is a risk factor for aspiration pneumonia. Tongue pressure measurement is a simple and noninvasive method for evaluating swallowing dysfunction. We have hypothesized that low tongue pressure may be able to predict pneumonia occurrence in acute stroke patients. Tongue pressure was measured using balloon-type equipment in 220 acute stroke patients. The modified Mann Assessment of Swallowing Ability (MASA) score was evaluated independently on the same day. Tongue pressure was measured every week thereafter. An improvement in tongue pressure was observed within the first 2 weeks. Receiver operating curve analysis was performed to determine the ability of tongue pressure to predict modified MASA score <95, which suggests swallowing dysfunction. The optimal cutoff for tongue pressure was 21.6 kPa (χ2 = 45.82, p<0.001, sensitivity 95.9%, specificity 91.8%, area under the curve = 0.97). The tongue pressure was significantly lower in patients with pneumonia than in those without pneumonia. Using a Cox proportional hazard model for pneumonia onset with a cutoff tongue pressure value of 21.6 kPa and adjustment for age, sex, and National Institutes of Health Stroke Scale score at admission, the tongue pressure had additional predictive power for pneumonia onset (hazard ratio, 7.95; 95% confidence interval, 2.09 to 52.11; p = 0.0013). In the group with low tongue pressure, 27 of 95 patients showed improvement of tongue pressure within 2 weeks. Pneumonia occurred frequently in patients without improvement of tongue pressure, but not in patients with improvement (31/68 and 2/27, p<0.001). Tongue pressure is a sensitive indicator for predicting pneumonia occurrence in acute stroke patients.