This article reviews human observations on pioglitazone and bladder cancer risk. The PROspective pioglitAzone Clinical Trial In macroVascular Events trial showed an imbalance in bladder cancer between users of pioglitazone and placebo (14 versus six cases, p = 0.069). However, after excluding bladder cancer probably ascribed to other etiology, a blind assessment concluded that the imbalance might not be related to pioglitazone. Epidemiologic studies conducted in the United States and France using insurance databases independently suggested that pioglitazone use for >2 years might confer a 20%–40% higher risk. Another study evaluating bladder cancer risk in diabetic patients using the National Health Insurance in Taiwan did not find any incident bladder cancer case among 422 pioglitazone users for a follow-up of up to 3 years. Because observational studies may suffer from selection and information bias, and inadequate adjustment for confounders may inflate the estimated risk, causal inference from these studies should be interpreted with caution. While investigating cancer risk associated with a medication, indication bias should also be attended, especially when the medication is used at a late stage of the disease. Because pioglitazone is usually a second or third line antidiabetic agent, the users are always characterized by older age, longer diabetes duration, poorer glycemic control, and higher rates of complications and comorbidities. Biased estimates will also result if these differences are not appropriately addressed in the analyses. Current evidence neither concludes nor excludes a causal role of pioglitazone on bladder cancer. Clinical trials aiming at evaluating the risk of cancer associated with a medication is not ethical and may not be expected to provide an answer on the issue of pioglitazone-related bladder cancer. However, a meta-analysis using all available clinical trials to compare the bladder cancer risk between pioglitazone and comparators will be helpful. Well-conducted epidemiologic observational studies are probably other options. Because bladder cancer incidence and their risk factors may differ significantly among different ethnicities, a clarification of such a link in different ethnicities is needed, better by using long-term databases with large and representative sample size and appropriate adjustment for confounders. Furthermore, the interactions with other comorbidities and concomitant medications should be addressed.