International audienceLarge and load-bearing bone defects are challenging to treat and cause pain and disfigurement. The design of efficacious bone scaffolds for the repair of such defects involves a range of length scales from the centimeter down to the micrometer-scale. Here, we assess the influence on bone regeneration of scaffold rod spacing (>300 μm) and microporosity (<50 μm), as well as the combination of different structures and materials in the same scaffold, i.e., at the millimeter scale. We use four single-domain scaffolds, microporous (MP) or nonmicroporous (NMP) and with either a “small” or “large” rod spacing. Multidomain scaffolds combine four regions corresponding to the macro- and microarchitectures of the single-domain scaffolds. The scaffolds are implanted in pig mandibles for 3 weeks and bone regeneration is assessed by measuring the average bone volume fraction, , the bone distribution and the trabecular thickness from micro-CT data. For the single-domain scaffolds, was 45 ± 3% for MP-small, 39 ± 2% for MP-large, 25 ± 2% for NMP-small, and 25 ± 2% for NMP-large. MP scaffolds have significantly higher and a more uniform bone distribution compared to NMP, regardless of rod spacing. The average trabecular thickness is significantly larger in MP compared to NMP, and in “large” compared to “small” scaffolds. Microporosity affects trabecular thickness throughout the scaffold, while rod spacing affects it only at the scaffold periphery. In multidomain scaffolds, MP-large and NMP-large domains have similar as compared to their respective single-domain counterparts. These results suggest that combining different architectures into one scaffold conserves the properties of each domain. Hence, bone growth and morphology can be tailored by controlling scaffold architecture from the millimeter down to the micrometer level. This will allow the customization of scaffold designs for the treatment of large and load-bearing bone defects