Background and aims. Positive below-ground interactions (facilitation) should be more pronounced when resources limit crop growth, according to the stress-gradient hypothesis. Our aim was to test this hypothesis for intercropped durum wheat and faba bean along a P-fertilizer gradient. Methods. A field experiment was conducted in a long-term P-fertilizer trial with three rates of P-fertilization (No, Low and High P). Microbial biomass was assessed by chloroform fumigation-extraction. Quantitative PCR was applied to evaluate the abundance of relevant microbial groups. Results. Phosphorus availability and microbial biomass systematically increased in the rhizosphere compared to bulk soil. P-fertilization resulted in higher abundance of targeted bacterial phyla, whole bacterial and fungal communities, and depressed mycorrhizal colonization of durum wheat, but not faba bean. Microbial biomass carbon significantly increased in the rhizosphere only in P-fertilized treatments, pointing to P limitation of microbial communities. Intercropping yielded a significant effect on rhizosphere microbial properties only at High P. Microbial biomass P increased in the rhizosphere of intercropped faba bean only at No P level, and was thus the sole finding supporting the stress-gradient hypothesis. Conclusions. P-fertilization was the main driver of microbial communities in this field trial, and P-fertilizer application modulated the species-specific effect in the intercrop. Plant performance did not validate the stress-gradient hypothesis as positive plant-plant interactions occurred regardless of the level of P-fertilization.