NC_Ar+NH3 is a Fe/N/C electrocatalyst developed by our group to reduce oxygen. It is obtained from the pyrolysis in Ar, then in NH3, of a mixture of ZIF-8 (a metal-organic-framework) and the complex made between ironII acetate and 1,10 phenantroline. In terms of activity and performance, NC_Ar+NH3 is becoming a serious contender to Pt, but it is still unstable in a H2/O2 PEM fuel cell. Its current density recorded at 0.6 V and 80 °C decreases, first quickly during about 15 h, then more slowly for several tens of hours. The aim of this work is to verify if iron in the catalyst could be at the origin of the first rapid decay of NC_Ar+NH3 through a Fenton reaction with some H2O2 generated by an incomplete reduction of O2 in fuel cell. To do so, the strategy was simple: produce a MOF_Ar+NH3 catalyst using the same ZIF-8, but without deliberate addition of iron precursor, and compare its stability to that of NC_Ar+NH3. However, this has been impossible to achieve since there is a native iron impurity in ZIF-8 leading to an increase of Fe content after each pyrolysis step. In order to circumvent this problem, we produced several MOF_Ar+NH3 (t) catalysts, varying (t), the pyrolysis time in NH3 (and therefore also the Fe content). This enabled us to determine, by extrapolation to 50 ppm Fe on a log i vs log Fe content, the initial current density at 0.6 V of a catalyst (MOF_CNx_Ar+NH3) for which the current density would essentially be attributable to CNx catalytic sites. From the similarity of the normalized instability curves for MOF_CNx_Ar+NH3, all MOF_Ar+NH3 (t) and NC_Ar+NH3 catalysts, we are able to conclude that neither iron (through a Fenton reaction) nor H2O2 alone are responsible for the first rapid decay of these Fe/N/C catalysts in fuel cells, but that a slow electro-oxidation of the carbonaceous support of all NC_Ar+NH3 and MOF_Ar+NH3(t) catalysts, occurring in about 15 h in H2/O2 fuel cell, is transforming the initially hydrophobic catalyst layers into hydrophilic ones. It is proposed that this phenomenon, causing micropore flooding, is at the origin of the first quick decay at 0.6 V of all the catalysts studied in this work. The slow electro-oxidation of the carbonaceous support during the durability test at 0.6 V also affects the mass activity of the catalysts. This was demonstrated for NC_Ar+NH3. We also determined what would be the initial polarization curve, the initial maximum power of MOF_CNx_Ar+NH3 (0.150 W cm−2; 22% of that of NC_Ar+NH3), the initial mass activity at 0.9 V of MOF_CNx_Ar+NH3 (0.3 A g−1; 3% of that of NC_Ar+NH3) and its initial Tafel slope (56 mV decade−1). The Tafel slope similarity for MOF_CNx_Ar+NH3 and for NC_Ar+NH3 (60 mV decade−1) indicates that the ORR kinetics on both types of catalysts, and therefore on CNx and FeNx catalytic sites, is also similar.