Introduction: It is well known that patients with objective response to pembrolizumab have a durable duration of response, leading to favorable survival outcomes. We investigated the possibility of predicting the objective response with concise indicators obtained from daily clinical practice. Methods: In our multi-institutional cohort, 220 platinum-refractory metastatic urothelial carcinoma (mUC) patients treated with pembrolizumab for at least 6 weeks with complete information of objective response were investigated. Results: The median follow-up was 7.3 months, and 119 patients deceased during the follow-up. A multivariate logistic regression analysis exhibited two independent variables predicting the objective response, including the neutrophil-lymphocyte ratio (NLR) change at 6 weeks of treatment and liver metastasis. We proposed a risk group using these two indicators. Patients with no predictive indicators/one of those were assigned to favorable (42%)/intermittent (47%) risk groups. Patients with both indicators were assigned to poor risk (11%). Notably, the objective response rate was well delineated in 41%, 25%, and 0% for favorable-, intermediate-, and poor-risk groups, respectively (p < 0.001). Distinct overall survival (OS) between the risk groups was also confirmed with the median OS of 14.1, 11.7, and 4.2 months in favorable-, intermediate-, and poor-risk groups, respectively. Conclusions: At the 6 weeks of the pembrolizumab treatment, our risk model predicts the objective response rate precisely. Notably, those classified as “poor risk” – marked by liver metastasis and an increased NLR – should be considered for alternative therapy with a different mode of action, highlighting a critical decision point in treatment optimization.