The development of degradable polymer scaffolds is a key issue in bone regeneration. Poly(D, L-lactide) (PDLLA) and its derivatives have usually been applied to the construction of degradable scaffolds, but these scaffolds had problems with acidic degradation products and quick loss of mechanic strength during the later degradation, which usually led to scaffold collapse and cavity formation because of the slower rate of bone regeneration. In the present paper, a polysaccharide derivative, agarose acetate (AGA), was synthesized and a novel porous AGA scaffold was successfully developed through a salt-leaching process. The AGA scaffold had over 90% porosity without swelling in water, and compared to collapse and acidic products of PDLLA scaffold during degradation, the AGA scaffold maintained a stable morphology and a nearly neutral pH value over 18 months' degradation in PBS. A bone mesenchymal stem cells (BMSCs) adhesion and proliferation experiment showed that more cells adhered to the AGA scaffold than to the PDLLA scaffold. A subcutaneous implant test showed that the AGA scaffold slowly degraded and did not cause an inflammatory response surrounding the implantation lesion site. AGA scaffold was implanted into femur defects in New Zealand white rabbits to test its in vivo performance. Results indicated that the AGA scaffold accelerated the process of bone regeneration compared to the PDLLA group and, with time, new bone was formed from the margin toward the center of the scaffolds, and the scaffold left in place retained its porous structure without collapsing. Meanwhile, the AGA scaffold showed a low degradation rate and kept its shape during the in vivo degradation compared to the PDLLA scaffold. This performance could have the benefit of integrated regenerative bone being formed instead of cavities due to the quickly degraded scaffold disappearing. These results demonstrate that the AGA scaffold has significant potential in bone regeneration applications.