Genitourinary (GU) cancers are among the most common malignant diseases in men. Rapid screening is the key to GU cancer management for early diagnosis and treatment. Urine is a highly accessible specimen type and urine metabolic fingerprints (UMFs) reflect underlying metabolite signatures of GU cancers. Herein, rapid screening of GU cancers is performed using high-throughput extraction of UMFs by mass spectrometry and efficient recognition by machine learning (ML). GU cancer patients can be distinguished with an accuracy of 90.1%. Besides, key biomarkers such as citric acid were found remarkably upregulated in cancer groups, indicating the dysregulated pathways. This approach highlights the potential role of ML in clinical application and demonstrates the expanding utility of UMFs in disease screening.