1. 1. The article presents the principal results of a four-year pollen census carried out with a Hirst automatic volumetric spore trap. 2. 2. Direct volumetric estimates have been made of the pollen incidence of the various types of pollen concerned; the potential value of such estimates in the study of inhalant allergy needs no emphasis. 3. 3. The survey has confirmed the general conclusions (based on a much longer gravity slide census) on the qualitative and (with the exceptions stated below) the quantitative botanical composition of the pollen cloud over South Wales, on the phenology and (with the same exceptions) the relative abundance of its constituent types, and on their variation from year to year. 4. 4. Use of an efficient volumetric trap has made possible a more correct assessment of the numerical abundance of certain types (notably Urtica) whose relative importance, by reason of their small size, had previously been underestimated. 5. 5. Attention is drawn to especially wide variations in the annual incidence of certain principal types and to the need to consider such variations when the effects of treatment are assessed.