It is now widely believed that there are two major pathways for urothelial carcinogenesis. One pathway usually involves mutation of FGF receptor 3 and gives rise to low-grade papillary tumors that frequently recur but seldom invade. By contrast, high-grade urothelial malignancies, including high-grade papillary urothelial carcinoma and urothelial carcinoma in situ (CIS) usually exhibit deletions or mutations of TP53. Urothelial CIS is the most likely precursor of high-grade invasive bladder cancer. It is a 'flat lesion' that may be relatively inconspicuous at cystoscopy, or even endoscopically undetectable. The clinical hazards associated with this elusive and biologically dangerous neoplasm have been increasingly well documented since the original studies by Melicow in 1952. Primary or secondary urothelial dysplasia is even more challenging to detect and diagnose than CIS. It is theorized that dysplasia may antedate the onset of CIS, but support for the putative progression of dysplasia to CIS is found in fewer than 20% of cases. Since many benign urothelial changes may resemble CIS at cystoscopy, in biopsies and even with molecular profiling, care must be exercised when making a diagnosis of CIS. For patients whose screening tests are worrisome for the presence of premalignant urothelial disease, newer bladder imaging modalities, including Raman spectral imaging and optical coherence tomography, may enable improved biopsy site selection. In this article, we discuss the above-noted topics, as well as other related issues, such as the possible role of papillary urothelial hyperplasia as a preneoplastic lesion and the roles of cancer stem cells and field cancerization in urothelial carcinogenesis.