The study presents the author's experience of 15 years of electron microscopic study of 339 resected respiratory tract tumours, an unusually large group. A dual purpose was pursued: a) to present a systematic analysis of carcinomas which prevailed in the material, and b) to interpret some of the unusual or poorly studied phenomena found in some tumours. The author has been using electron microscopy to study carcinomas since 1973 as a corrective factor giving more precision to routine classification (according to WHO, 1981), on which modern diversified therapy depends. He presents more evidence of the phenotypic heterogeneity of carcinomas, puts aside the histogenetic approach and supports the idea of a common genotypical source of all the programmes of tumour cell differentiation. As regards individual tumours studied by electron microscopy, the author introduced the first ever observation of extrapulmonary neuroendocrine laryngeal carcinoma verified at post mortem examination, a third observation of filamentous bodies in four bronchial carcinoids, and the first observation of a flat epitheloid hemangioma of the larynx. Another case, one of intravascular bronchioloalveolar tumour, is unique for the presence of cytoplasmic crystalline inclusions and leptomeric fibrils which used to be associated with contractile proteins. In a case of bronchioloalveolar carcinoma resembling jaagsiekte, unusual paracrystalline inclusions are demonstrated in a second ever observation, and diagnosed as viral metabolites. Radial cisternae, a peculiar variety of annulate lamellae, found in a lung adenocarcinoma receive closer scruitny because altogether five such cases are on record (two more tumours in the author's archives and another two taken from illustrations to other authors' reports).