Archived wheat (Triticum aestivum L.) grain and straw, and soil samples from the control plot of the Rothamsted Broadbalk Experiment, located in southeastern England and established in 1843, were used to investigate the effects of dramatically changing SO2 pollution inputs on the concentrations and stable isotope ratios (delta34S) of S in the samples. Representative coal samples from UK major coal fields were also determined for delta34S. Concentrations of S showed no clear trends in either grain or straw over the 155 years from 1845 to 1999. However, grain and straw delta34S decreased rapidly from 6 to 7/1000 in 1845 to -2 to -5/1000 in the early 1970s, and since then have increased to 0.5 to 2/1000 in the late 1990s. This pattern mirrored the trend of UK SO2 emissions over the 155 years. Both grain and straw delta34S correlated strongly and negatively with UK SO2 emissions (R2 > 0.89), but the relationships were different for the pre- and post-1970 data sets. Soil delta34S also decreased considerably, from 8.2/1000 in 1865 to 3.7 to 4.5/1000 during 1965-1999. A negative delta34S value was inferred for the anthropogenic S deposited at the experimental site before 1970, and further confirmed by negative delta34S values (-6 to -10/1000) found in the coal samples from southeastern England and southern Wales. Based on the S isotope ratios, we estimated that anthropogenic S contributed 62 to 78% of the S uptake by wheat at the peak of SO2 emissions, and accounted for 28 to 37% of the topsoil S in 1965.