There have been few attempts to quantify the causes for annual variations in the frequency of pollen in the air. The amounts of pollen deposited annually in Tauber traps exposed in a mixed deciduous forest in southwestern Denmark 1967–76 were measured. The variations in the frequency of pollen from the local trees showed distinct biennial patterns. Flowering intensity in years following a year with low flowering was correlated with precipitation in April of the preceding year (negative,BetulaandAbuts) or average summer temperature also of the preceding year (QuercusandFagus). The pollen frequencies of grasses, herbaceous plants andCallunawere correlated with the precipitation of seasons that preceded the flowering. Years with very high pollen frequencies (1968 and 1969) must have been unpleasant for hay fever patients. Such severe pollen seasons can be predicted with fair accuracy.