Gas exchange was measured from 1 month before the onset of anthesis until the end of grain filling in field-grown spring wheat, Triticum aestivum L., cv. Vinjett, in southern Sweden. Two g ( s ) models were parameterised using these data: one Jarvis-type multiplicative g ( s ) model (J-model), and one combined stomatal-photosynthesis model (L-model). In addition, the multiplicative g ( s ) model parameterisation for wheat used within the European Monitoring and Evaluation Programme (EMEP-model) was tested and evaluated. The J-model performed well (R (2)=0.77), with no systematic pattern of the residuals plotted against the driving variables. The L-model explained a larger proportion of the variation in g ( s ) data when observations of A (n) were used as input data (R (2)=0.71) compared to when A (n) was modelled (R (2)=0.53). In both cases there was a systematic model failure, with g (s) being over- and underestimated before and after anthesis, respectively. This pattern was caused by the non-parallel changes in g ( s ) and A (n) during plant phenological development, with A (n) both peaking and starting to decline earlier as compared to g ( s ). The EMEP-model accounted for 41% of the variation in g ( s ) data, with g ( s ) being underestimated after anthesis. We conclude that, under the climatic conditions prevailing in southern Scandinavia, the performance of the combined stomatal-photosynthesis approach is hampered by the non-parallel changes in g ( s ) and A (n), and that the phenology function of the EMEP-model, having a sharp local maximum at anthesis, should be replaced by a function with a broad non-limiting period after anthesis.