Taxon discovery underlies many studies in evolutionary biology, including biodiversity and conservation biology. Synonymy has been recognized as an issue, and as many as 30–60% of named species later turn out to be invalid as a result of synonymy or other errors in taxonomic practice. This error level cannot be ignored, because users of taxon lists do not know whether their data sets are clean or riddled with erroneous taxa. A year-by-year study of a large clade, Dinosauria, comprising over 1000 taxa, reveals how systematists have worked. The group has been subject to heavy review and revision over the decades, and the error rate is about 40% at generic level and 50% at species level. The naming of new species and genera of dinosaurs is proportional to the number of people at work in the field. But the number of valid new dinosaurian taxa depends mainly on the discovery of new territory, particularly new sedimentary basins, as well as the number of paleontologists. Error rates are highest (>50%) for dinosaurs from Europe; less well studied continents show lower totals of taxa, exponential discovery curves, and lower synonymy rates. The most prolific author of new dinosaur names was Othniel Marsh, who named 80 species, closely followed by Friedrich von Huene (71) and Edward Cope (64), but the “success rate” (proportion of dinosaurs named that are still regarded as valid) was low (0.14–0.29) for these earlier authors, and it appears to improve through time, partly a reflection of reduction in revision time, but mainly because modern workers base their new taxa on more complete specimens. If only 50% of species are valid, evolutionary biologists and conservationists must exercise care in their use of unrevised taxon lists.