Intercropping systems consist in growing two or more species simultaneously in the same field for at least part of their growth cycle. In situations of limiting nutrient availability, acquisition of resources in intercrops is generally improved compared to sole crops, as a result of positive interactions between the intercropped plant species, i.e. either niche complementarity or facilitation. Evidence has recently accumulated that intercropped cereals and grain legumes promote the use of soil P resources, though few reports derived from field trials. We investigated whether (i) intercropping of durum wheat with pea or faba bean does improve soil P acquisition; (ii) facilitative interactions can be detected, and (iii) soil P availability does influence such effects. We conducted two 1-year experiments in a long term P fertilizer field trial with a gradient of Olsen-P soil content from <5 to 30 mg/kg (Auzeville, SW France; 2009 and 2011). Aboveground parameters such as growth, productivity, nutritional status (N and P), fixed N ratios and grain quality were regularly monitored on intercrops and sole crops. Relative total biomass was always higher in intercrops vs. sole crops as illustrated by LER>1, except in the lower P soil content, where faba bean was outcompeted by wheat in the later stages. P export was increased in intercrops vs. sole crops at the legume flowering stage, while it decreased afterwards to similar amounts mainly due to faba bean suppression. Our data suggest positive plant-plant interactions (niche complementarity/facilitation) at earlier growth stages of wheat-legume intercrops that transiently resulted in a more efficient use of soil P resources but then receded until harvest.