N fixation is one of the most important microbially driven ecosystem processes on Earth, allowing N to enter the soil from the atmosphere, and regulating plant productivity. A question that remains to be answered is whether such a fundamental process would still be that important in an over-fertilized world, as the long-term effects of fertilization on N fixation and associated diazotrophic communities remain to be tested. Here, we used a 35-year fertilization experiment, and investigated the changes in N fixation rates and the diazotrophic community in response to long-term inorganic and organic fertilization. It was found that N fixation was drastically reduced (dropped by 50%) after almost four decades of fertilization. Our results further indicated that functionality losses were associated with reductions in the relative abundance of keystone and phylogenetically clustered N fixers such as Geobacter spp. Our work suggests that long-term fertilization might have selected against N fixation and specific groups of N fixers. Our study provides solid evidence that N fixation and certain groups of diazotrophic taxa will be largely suppressed in a more and more fertilized world, with implications for soil biodiversity and ecosystem functions.