As part of a long-term open-air fumigation experiment, Scots pine was exposed continuously to SO,^ pollution following a predetermined pattern of hourly mean values based upon monitoring data from a site in central England. Altbougb annual mean concentrations did not exceed 22 ppb, up to 20 "o of trees developed foliar necrosis in tbree consecutive growing seasons similar to publisbed descriptions of acute SOj damage. Tbe number of trees affected increased witb SOj concentration, and tbere was a significant tendency for tbe same trees to be affected in consecutive years, implying tbat sensitivity to these levels was genetically based. As a group, trees showing necrotic symptoms were not significantly retarded in tbeir growtb, altbougb tbeir 1990 budburst was significantly delayed and tbe few trees wbicb lost over 50% of tbeir foliage were visibly stunted. Fumigation of potted trees for sbort periods in chambers witb up to 1000 ppb SO.^ failed to produce any comparable effects. Regression analyses between tbe number of plants affected eacb year and a range of gas concentration parameters sbowed mean concentrations during a critical period in May to be most closely related to tbe development of symptoms, altbougb no effects became visible until approximately 5 wk later. Tbe start date for tbe critical period difFered between years, and for tbe two years for whicb budburst data were available it coincided witb tbe start of needle elongation. Tbe duration of the critical period lay in tbe range 6-14 d. Linear extrapolation from the data suggested tbe tbresbold mean SO^ concentration for visible injury to be in tbe range 6-8 ppb (measured during tbe critical period during needle expansion) altbougb tbis could underestimate tbe true value.