Current regulatory programs for drug safety are not designed to identify adverse events that have a long induction time or are rare, such as most cancers. Meta-analyses of randomized clinical trials of medications can sometimes provide information on shorter-term risk of common cancer types, though large observational studies with long follow-up are needed to examine most drug–cancer associations. Over the last few decades, a number of new methods have been developed to address several types of confounding and bias of particular concern in pharmacoepidemiology, and better data sources have become available. Of the approximately twenty medications with sufficient evidence to be classified by the International Agency for Research on Cancer (IARC) as human carcinogens, most are anti-neoplastic agents or immunosuppressants. Substantial data from studies in humans indicate that use of aspirin and other nonsteroidal anti-inflammatory drugs (NSAIDs) protects against colorectal cancer and possibly a number of other common cancers.