A total of 179 published cladograms of arthropods was tested for their congruence with the palaeontological record. Congruence for data sets including 510 tetrapod and 157 fish trees was also calculated for comparison. Results provide support for the hypothesis that the fossil record of arthropods is less complete (at levels of stratigraphic and taxonomic resolution pertinent to most macroevolutionary studies) than the records of several other taxa. Differences in tree size, shape, balance and the distribution of first stratigraphic appearances (potential sources of bias) were controlled for by various randomization procedures. Most measures of congruence for arthropods were statistically poorer than those for other groups, even when the sample was divided into broad temporal bins. Many of the most robust and widely reproduced sister groupings of arthropods are attended by ghost ranges of many tens or hundreds of millions of years. Fossils of a number of presently very diverse and abundant arthropod groups are conspicuously absent from the record, despite many spectacular examples of the detailed preservation of others. There is probably no single reason for the apparently poor record of arthropods. Low preservation potential, small size of individuals, small numbers of individuals, and restricted palaeobiogeography almost certainly play a role in particular cases.