Field grown spring wheat (Triticum aestivum) cv. Dragon, with artificially altered source/sink ratios, was exposed to ozone in open-top chambers in order to test whether the ozone sensitivity was affected by a shift in the source/sink relationship. The source and sink were manipulated by removing the flag leaf or the upper part of the ear, respectively. In 1995, three different chamber treatments were used: non-filtered air without extra ozone (NF), with ozone added before anthesis (NF + pre) or with ozone added during and after anthesis (NF + post). The ozone exposure during the treatment periods was 2417 and 2508 nmol mol(-1) hours (AOT40) in NF + pre and NF + post, respectively. In 1997, the plants were exposed to filtered air (F) or to non-filtered air without extra ozone (NF) or with three different levels of ozone added (NF1 +, NF2 + and NF3 +). The ozone exposure expressed as AOT40 was 0, 13, 1924, 5881 and 10 375 nmol mol(-1) hours in F, NF, NF1 +, NF2 + and NF3 +, respectively. The dry weight per grain and the amount of total non-structural carbohydrates and nitrogen per grain were reduced by ozone exposure in 1997. In 1995, the dry weight per grain in the source-manipulated shoots was reduced by ozone, especially when the exposure was conducted during and after anthesis. The ozone effect in 1997 was smaller in sink-manipulated shoots than in source-manipulated and non-manipulated shoots. It is concluded that the ozone sensitivity of wheat is higher when the grain growth is source-limited. It is also concluded that the ozone sensitivity of wheat is higher during and after anthesis than before anthesis, although source-limited wheat plants may be sensitive also to early ozone exposure. (C) 2001 Elsevier Science B.V. All rights reserved.