This account is a generic monograph of a medium-sized (494 species) pantropical plant family. Its rank, circumscription and relationships to other families are examined. An attempt is made to provide an objective and repeatable taxonomy and to formulate evolutionary hypotheses consistent with the facts presented. The monograph is also used to test the general application of certain relatively new and still controversial theoretical approaches to descriptive and evolutionary biology, notably numerical taxonomy, cladistics and vicariance biogeography. The practical value of some potentially useful sources of taxonomic information, especially electron microscopy and phytochemical analysis, is also examined. This is possibly the first time that non-traditional approaches to taxonomy have been so comprehensively evaluated within the framework of a worldwide monograph. Based mainly on original investigations, the morphology, anatomy, reproductive biology, distribution and ecology of each of the 17 genera are described and illustrated. Published works by others on chemistry, embryology and chromosome cytology are summarized. The delimitation and arrangement of the genera included here depart somewhat from those of earlier systems. Most of the new taxa have been formally published elsewhere but the following are proposed here for the first time: tribus Chrysobalaneae, tribus Parinarieae Prance & F. White, tribus Couepieae Prance & F. White, tribus Hirtelleae Prance & F. White, Licania subgenus Angelesia (Korth.) Prance & F. White and Atuna latifrons (Kosterm.) Prance & F. White. All Chrysobalanaceae are woody. They vary greatly in size but are uniform in their vegetative architecture. By contrast, the structure of the inflorescence, flower and fruit is much more diverse; detailed examination has revealed a multitude of previously overlooked characters of classificatory importance. Individually these characters are small, but they are well correlated and belong to character complexes of high diagnostic value which are also functional syndromes related to pollination, dispersal and germination. Each genus is unique in the precise way in which its stigmatic and polliniferous surfaces are presented to pollen vectors and in the structure of its fruit, although within a genus there may be more than one principal pollinator or disperser. Except for a few anomalous species, each genus is distinct in its general appearance. Membership of a genus can be detected by intuitive perception, based on rapid inspection `by eye' of herbarium specimens with flowers. This is possible because of the underlying structural uniformity of the inflorescence and the similar proportions of the flower in all congeneric species. For two genera, Couepia and Hirtella, their objectivity and ease of identification by rapid visual means was demonstrated by simple experiments involving colleagues with little experience of taxonomy and no previous knowledge of the group. The characteristics these `subjects' had subconsciously used were then explicitly formulated in taxonomic terms and their individual diagnostic values calculated. Some methods of numerical taxonomy were found to be useful, although less so than their advocates frequently claim. This is because of problems associated with the detection, selection and weighting of characters. Attempts to apply cladistic analysis to the Chrysobalanaceae were largely unsuccessful because of parallelism and the difficulty of identifying sister groups. The occurrence and extent of parallelism was shown by using a `principle of discordant characters', which is defined and discussed. The ideas underlying it are widely known but the principle itself does not seem to have been previously used as an analytical tool in taxonomic research. In the Chrysobalanaceae, parallelism is so rife that convincing cladograms cannot be constructed; for this reason, among others, the methods of vicariance biogeography are inappropriate. Aspects of evolution other than branching patterns, however, are more relevant to general biology, and monographic studies should help to identify them. Examples from Chrysobalanaceae include: (a) the response in growth-form to geomorphological and climatic change; (b) the biological consequences of dispersal across faunistic barriers; and (c) the evolution of ocean-dispersed, obligate strand plants. It is concluded that monographers cannot afford to ignore fine structure and chemistry, though there is no evidence that they are more fundamental in practical taxonomy than other sources of information. The Chrysobalanaceae and other families show that parallelism is as likely to occur in fine structure and chemistry (at least of secondary compounds) as in gross morphology, and it is much more expensive to detect. At present, for tropical plants, some currently fashionable approaches to taxonomy are of limited relevance. It is much more important to develop objective methods of visual analysis for dealing with the copious material that monographic work entails. It is equally necessary to understand the central role of ecology in biology. Without ecological understanding taxonomy is lifeless and dull, and taxonomically important information and interesting evolutionary situations go undetected.