My Ph.D. research is based on nearly three years of fieldwork in Tanzania, from December 1978 to September 1981. It attempts to deduce the effects of certain ecological and biogeographic-factors on the structure and species composition of forest bird communities. My principal study areas were in the Usambara Mountains in north-eastern Tanzania, though other forests were visited. My research concentrated on three main aspects. 1. Factors affecting forest bird distribution. I compared the forest avifauna of the Usambaras with those of neighbouring forest areas to determine some of the factors that affect the distribution of forest birds. Altitude appears to have the most important influence on bird distribution and many species are either exclusively lowland or exclusively montane. A few species are restricted to narrow altitudinal bands at intermediate levels. Rainfall is another important factor. Forests in wetter areas tend to have rich avifaunas and several species are restricted to such moist forests. It seems likely that historical factors are also important. Localities in which forests are believed to have survived throughout the driest periods of the Pleistocene are the richest in species today. It appears that such localities often coincide with the areas of highest rainfall. Other factors, such as degree of isolation, vegetation sructure and inter-specific competition, have a less pronounced, though probably real, effect on the avifaunas. Latitude and forest area, however, have no noticeable influence. 2. Factors influencing avian altitudinal zonation. Birds are presumably not affected by altitude so much as by other factors which are themselves influenced by altitude. The Usambara Mountains proved to be an ideal study area for the altitudinal zonation of forest birds since a wide altitudinal span of easily accessible forest survives from only 150m up to over 2250m a.s.l. Birds were counted by means of mist-netting and line transect censuses. If the Usambara forest avifauna is analysed in terms of the presence or absence of species at each altitude, two major "breaks" are detected at 800m and 1650m respectively. Between these two altitudes there are large numbers of both lowland and montane species. In terms of the abundance of species, there is a major break at 450m, above which the avifauna changes in a continuous cline. Several factors probably contribute to this altitudinal zonation. Perhaps the most important is temperature. Many lowland forest species occur at high altitudes only on the forest edge and avoid the cold forest interior. There is a pariculariy noticeable "cut-off" altitude at 1650m, above which many insectivorous species are not found. This is the lowest altitude for the occurrence of cold season frosts. There is also some evidence of inter-specific competition with several pairs of lowland and montane species replacing each other altitudinally. Analyses of altitudinal zonation are greatly complicated by the seasonal vertical movements of forest birds, which I discovered to be far more extensive in eastern Tanzania than has hithertoo been realised. Many species breed at high altitudes in the hot season, moving to lower levels during the colder months. Also many frugivorous, gramrnivorous and nectivorous species are liable to erratic, apparently aseasonal movements in search for food. As a result care has had to be exercised in the interpretation of census results. 3. Comparisons between forest and non-forest bird communities. I also compared the avifaunas of forest and non-forest habitats in the Usambaras. On the Mazumbai Estate, a tea estate at 1500m in the West Usambaras, I had a number of study sites in several different habitat types. The birds were counted by line transect censuses and mist-netting and the habitat structure was measured by estimating the height of the vegetation layers. I also attempted to measure the ecological position of each study species by measuring ten habitat parameters in the territories of breeding birds. The results of these studies suggested that the rigid division of birds into forest and non-forest species, which many ornithologists in Africa have accepted, is simplistic. Many forest species are very conservative and unable to withstand too much interference with their habitat (and similarly many non-forest species avoid areas where there are trees). Several other species, however, are not easy to categorise and some birds characteristic of undisturbed forests are able to survive in some very degraded habitats. The rare and endemic birds in the Usambaras are all forest species and most tend to be among those less tolerant of habitat change. For this reason considerable concern has been ex pressed about the conservation of the Usambara forests. As a result of my research I have been able to draw up a conservation management plan. Forests also have important economic benefits, particularly in the preservation of water catchments and the prevention of soil erosion. Conclusion. In my discussion I have noted that tropical forest bird communities are particularly complex and require long field studies before trends and patterns can be clearly detected. In the past too many conclusions have been drawn from field studies lasting only weeks or months. The forest bird communities of eastern Tanzania may not be typical of those of the tropics as a whole. They appear to differ from those in West Africa and South America in at least three respects: (i). Forest area has little affect on the number of bird species. (ii). Interspecific corn petition does not appear to be particularly important. (iii). No species appears to require primary unexploited forest. I have suggested that these differences are due to the relatively impoverished forest avifauna in eastern East Africa. It seems probable that these forests lost many bird species during the dry periods of the Pleistocene and those that survive today are more adaptable and tolerant of reduction in forest area and exploitation of their habitat.