Abstract This study investigated the effect of the distribution of aged binder from reclaimed asphalt pavement (RAP) on the fracture tolerance of recycled asphalt mixtures. Two RAP sources (one coarse, one fine), one virgin polymer-modified asphalt binder, and four RAP contents (0 %, 20 %, 30 %, and 40 %) were used to produce a total of seven mixtures. Fracture tolerance was assessed in terms of fracture energy density (FED) at three levels: binder, interstitial component (IC) (composed of effective binder and fine aggregate fraction), and mixture. Binder fracture energy tests were conducted to measure the FED of blends of virgin and recovered RAP binder. IC direct tension tests were employed to determine the IC FED. Finally, Superpave indirect tensile fracture tests were performed to obtain the FED of RAP mixtures. FED decreased with increasing RAP content at the three levels (binder, IC, and mixture). More importantly, IC and mixture exhibited almost identical FED reductions in relative terms, while the reduction was less significant at binder level. This finding indicates that the full blending scenario artificially created at the binder level did not occur at IC and mixture levels. Furthermore, the coarse RAP yielded greater FED than the fine RAP, which substantiates that RAP gradation controls the distribution of RAP binder in recycled mixtures. Consequently, the lower content of aged RAP binder in the IC portion for mixtures with the coarser RAP resulted in better fracture tolerance. The application of a newly developed guideline for determining the maximum allowable RAP content in recycled mixtures confirmed that the coarser RAP used in this study allows for a greater RAP content than the finer RAP despite the stiffer RAP binder.