BACKGROUND: Immune checkpoint inhibitors (ICIs) improved survival in patients with locally advanced or metastatic urothelial carcinoma (la/mUC). Patient-reported symptoms in this context were poorly studied. The study aimed to compare symptom severity between patients and clinicians. METHODOLOGY: The secondary analysis of the AMI clinical trial comparing changes in the gut microbiota in patients with la/mUC treated with pembrolizumab was conducted in nine French centers. Secondary endpoints were expected in this prospective study. Patient-Reported Outcome-Common Terminology Criteria for Adverse Events (PRO-CTCAE) and CTCAE were assessed respectively by patients and clinicians before pembrolizumab initiation, and at each treatment visit until treatment cycle 12. Agreement in severity between clinicians and patients for grade ≥ 3 symptoms was calculated with Cohen's kappa coefficient. The toxicity index was generated for CTCAE and PRO-CTCAE to assess discordance in a longitudinal manner. The Wilcoxon test was used to compare clinicians' and patients' toxicity index and symptom severity frequencies. RESULTS: Thirty-nine patients were included (M/F sex ratio: 2.5) from December 2020 to March 2022. PRO-CTCAE baseline completion rate was 77.5%. Cohen's kappa coefficient ranged from -0.017 (95% confidence interval (CI), [-0.039, 0.005]) for numbness/tingling to 0.161 (95% CI, [0.045, 0.276]) for fatigue. The patient self-rated symptom toxicity index was > 2 for all symptoms compared to ≤ 0.62 (fatigue) when assessed by clinicians in longitudinal reporting of symptom frequency and severity with a p value < 0.001. The three most commonly reported symptoms by patients and clinicians, respectively, were: Fatigue 53.3% versus 23.4%, generalized pain 42.4% versus 16.5%, and insomnia 41.1% versus 9.5%. Symptom frequency reports between clinicians and patients were statistically different (p < 0.009). CONCLUSIONS: Symptom severity assessment showed discordance between patients and physicians. Clinicians reported fewer symptoms and graded them less severely than patients. PROs should be used to accurately reflect patient experience. TRIAL REGISTRATION: ClinicalTrials.gov identifier: NCT04566029.