It is often claimed that agroforestry could increase the total productivity per land unit compared to monocropping systems. The aim of this study was to evaluate, in a sub-humid Mediterranean climate, the behavior of the yield components, phenology, LAI and NDVI of durum wheat in an alley-cropping system. Our hypothesis was that the microclimate changes in agroforestry could change the devlelopment and yield of cereals. Two different experiments were carried out: in 2015 under 16-year old poplars in East–West lines and in 2016 under 21-year-old ash trees in North–South lines. In each experiment, 12 genotypes of durum wheat were sown. The grain yield was not significantly different in agroforestry and full sun conditions in 2015; however, both systems in this experiment had a particularly low yield (≈ 10% of the historical average yield of the plot). In 2016, the grain yield was significantly lower in agroforestry in comparison with full sun conditions. In both experiments, the most impacted yield component by agroforestry was the number of grains per spike. Similarly, in both experiments, the number of grains per spike was the only yield component impacted by the position within the alley inside agroforestry. Surprisingly, in 2016 the grain yield was higher in the West than in the center position of the alley. In both experiments, agroforestry delayed the maturity of the crop. The use of standard growing degree days was not sufficient to explain the difference in phenology between agroforestry and full sun conditions.