Strategies for plant resistance gene deployment aim to preserve their durability faced to highly adaptable fungal pathogens. While the pyramiding of resistance genes is often proposed as an effective way to increase their durability, molecular mechanisms by which the pathogen can overcome the resistance also are important aspects to take into account. Here we report a counterexample where pyramiding of two resistance genes of Brassica napus, Rlm3 and Rlm7, matching the Leptosphaeria maculans avirulence genes AvrLm3 and AvrLm4-7, respectively, favored the selection of double-virulent isolates. We previously demonstrated that the presence of a functional AvrLm4-7 gene in an isolate masks the Rlm3-AvrLm3 recognition. Rlm7 was massively deployed in France since 2004. L. maculans populations were surveyed on a large scale (>7600 isolates) during 20 years, and resistance gene deployment at the regional scale was determined. Mutations in isolates overcoming both resistance genes were analysed. All these data indicated that the simultaneous success of Rlm7, the deployment of varieties pyramiding Rlm3 and Rlm7, along with the decrease in areas cultivated with Rlm3 only, contributed to the success of virulent isolates toward Rlm7, and more recently to both Rlm3 and Rlm7. Experimental field assays validated that resistance gene alternation was a better strategy compared to pyramiding in this context. Our study also illustrated an unusually high sequence diversification of AvrLm3 and AvrLm4-7 under such a selection pressure, and identified a few regions of the AvrLm4-7 protein involved in both its recognition by Rlm7 and in its AvrLm3-Rlm3 masking ability.