The role of ear photosynthesis in grain filling was studied in a number of durum wheat (Triticum turgidum var durum L.) landraces and varieties from the Middle East, North Africa, and from the collections of ‘Institut National de la Recherche Agronomique’ (INRA, France) and ‘Centro International de Mejora de Maiz y Trigo’ (CIMMYT, Mexico). Plants were grown in the field in a Mediterranean climate. Flag leaves (blade plus sheath) and ears were kept in the dark from 1 week after anthesis to maturity which reduced grain weight by 22.4% and 59.0%, respectively. In a further experiment, the carbon isotope discrimination ratio (Δ) of ear bracts, awns and flag leaves was measured on samples taken at anthesis and on mature kernels. The mean value of Δ for the water soluble fraction of bracts (17.0‰) and awns (17.7‰) were lower than those of leaves (19.5‰) and fairly similar to those of kernels (17.4‰) averaged across all genotypes. Data indicate that most of the photosynthates in the grain come from ear parts and not from flag leaves. In addition, a higher water use efficiency (WUE) of ear parts than of the flag leaf is suggested by their lower Δ values. Gas exchange in ears and flag leaves was measured during grain filling. Averaged over all genotypes, CO2 diffusive conductance was about five times higher in the flag leaf than in the spike (with distal portions of awns outside the photosynthetic chamber) 2 weeks after anthesis. In absolute terms, the dark respiration rate (Rd) was greater than the net photosynthesis rate (Pn) by a factor of 1.74 in the spike, whereas Rd was much smaller, only 22.1, 65.7 and 24.8% of Pn in blade, sheath and awns, respectively. Data indicate that photosynthesis, and hence the water use efficiency (photosynthesis/transpiration), is greatly underestimated in ears because of the high rates of respiration which diminish the measured rates of net CO2 exchange. Results of 13C discrimination and gas exchange show that genotypes from North Africa have higher WUE than those from the Middle East. The high Rd values of ears as well as their low diffusive conductance suggest that CO2 from respiration may be used as source of carbon for ear photosynthesis. In the same way, the anatomy of glumes, for example, supports the role of bracts using internal CO2 as source of photosynthesis. In the first experiment, the Δ in mature grains from culms with darkened ears compared with control culms provided further evidence in support of this hypothesis. Thus, the Δ from kernels of control plants was 0.40 higher than that from ear-darkened plants, probably because of some degree of refixation (recycling) of respired CO2 in the grains.