Background Transurethral resection of bladder tumor (TURBT) is the standard treatment for non-muscle-invasive bladder cancer (NMIBC), but the high rates of residual tumor and recurrence remain major clinical challenges. This study aimed to evaluate the clinical utility of UroCAD urine test for detecting residual tumor after TURBT and predicting recurrence in high-grade bladder cancer.Methods In this prospective study, 38 patients with high-grade bladder cancer were divided into a single TURBT group (n = 23, followed ≥ 6 months for recurrence) and a second TURBT group (n = 15, repeat resection at 4–6 weeks). Urine samples were collected 7–14 days post-initial TURBT for UroCAD-based chromosomal instability (CIN) analysis. Diagnostic performance was assessed against pathology or recurrence.Results In the single TURBT group, UroCAD showed 80% sensitivity, 88% specificity, area under the ROC curve (AUC) 0.84, negative predictive value (NPV) 88%, and 85% accuracy for predicting recurrence. In the second TURBT group, sensitivity for residual tumor was 0%, specificity 92%, AUC 0.46, NPV 86%, and accuracy 80%. Marked CIN in chromosomes 1,3,7,17 was observed in the single TURBT group and pre-second TURBT subgroup, but significantly decreased after second TURBT. Recurrent copy number variations (CNVs) (5p,20q,10p,12q) were identified in the single TURBT group. Maximum CIN Z-values post-second TURBT were significantly lower than those in the single TURBT group (P < 0.05).Conclusion UroCAD-based CIN detection is a promising non-invasive tool for predicting recurrence in high-grade bladder cancer patients after single TURBT and monitoring treatment response. However, the extremely low sensitivity of UroCAD in detecting residual tumors before second TURBT requires further optimization through larger-scale studies.