To compare the gastrointestinal (GI) toxicities of various nonsteroidal anti-inflammatory drugs (NSAIDs), it is first necessary to dismantle the imprecise entity "NSAID gastropathy" into its component conditions, and understand their pathogenesis. In this article, toxicities are reviewed only in so far as they affect the upper GI tract, the site of approximately 80% of GI injuries caused by NSAIDs. The phenomena discussed are platelet dysfunction, causing hemorrhage from various lesions, including but not confined to ulcers; various mucosal injuries, ranging from small erosions to large ulcers; and "complications," such as bleeding or perforations, causing admission to hospital, that may arise from ulcers but may also arise from other lesions. Ulcers, when complicated, may either be those caused by an NSAID or "peptic" ulcers that preceded NSAID therapy (having a high prevalence in the population) and gave rise to complications resulting from NSAID effects on platelets, tissues, or biologic processes, for example, healing, necrosis/apoptosis, leukocyte adherence, vasoconstriction, or generation of free radicals. NSAIDs have been compared in various ways, including fecal blood loss, endoscopic lesion development, prospective preclinical cohort studies measuring perforations, ulcers, and bleeds, post-marketing surveillance studies, and studies of the incidence of serious adverse events in populations followed in large databases linked to each individual patient record with regard to drug consumption and outcome. All methods show considerable differences between NSAIDs. Modern studies on the relative toxicities of NSAIDs are summarized and reviewed, and a number of marketed and emerging drugs that appear less toxic than classic NSAIDs are identified.