Glossina, palpalis (R.-D.) was studied in three localities near Voinjama, in a forest region in north-west Liberia, where human trypanosomiasis persists only as sporadic cases. Sites were chosen with different degrees of contact between man, fly and game, in transitional vegetation on the Zeliba River, and in primary forest on the Lofa and Lawa Rivers. The general features, fauna and climate of these areas are described. Catching by nets was found to be preferable to the use of traps. Trap catches averaged about one-quarter of the corresponding net catches. The theoretical basis of ‘animal’ traps is too uncertain for their use in quantitative studies, and under forest conditions there are practical difficulties in siting them in optimum positions; they failed to catch at all at very low fly densities. Heavy rain and the dry season were unfavourable to the fly, particularly outside the forest canopy. A wet-season resting site on horizontal twigs near to the ground is described. The percentage of females in the total catch differed significantly between different habitats, but apparently was not indicative of the degree of hunger in the population. There were no great differences between the sexes in activity patterns, and activity was greatest during the apparently unfavourable climatic conditions of early afternoon. The willingness of the fly to penetrate unfavourable surroundings suggests that the clearing of forest at potential places of transmission is unlikely to afford an effective means of protection. There is some evidence that breeding was curtailed during the rains. Very few engorged flies were found. Fly numbers and distribution were not affected by normal game movements. There is no evidence that the fly was ever dependent on man for food. Where game was driven away the fly tended to follow it. Occasional human foci did not affect the number or distribution of the fly, and fly activity was not correlated with human activity. Intimate and prolonged contact between man and fly did not occur.