We describe two studies encompassing the iterative refinement of a polymer-based rhBMP-2 delivery system for bone tissue engineering. Firstly, we compared the bone-forming capacity of porous poly(D,L-lactic-co-glycolic acid) (PLGA) scaffolds produced by thermally induced phase separation (TIPS) with non-porous solvent cast poly(D,L-lactic acid) (PDLLA) used previously. Secondly, we examined the potential synergy between rhBMP-2 and local bisphosphonate in the PLGA scaffold system. In vivo ectopic bone formation studies were performed in C57BL6/J mice. Polymer scaffolds containing 0, 5, 10 or 20 µg rhBMP-2 were inserted into the dorsal musculature. At all rhBMP-2 doses, porous PLGA produced significantly higher bone volume (BV, mm3) than the solid PDLLA scaffolds. Next, porous PLGA scaffolds containing 10 µg rhBMP-2 ± 0.2, or 2 µg zoledronic acid (ZA) were inserted into the hind-limb musculature. Co-delivery of local 10 µg rhBMP-2/2 µg ZA significantly augmented bone formation compared with rhBMP-2 alone (400 % BV increase, p < 0.01). Hydroxyapatite microparticle (HAp) addition (2 % w/w) to the 10 µg rhBMP-2/0.2 µg ZA group increased BV (200 %, p < 0.01). We propose that this was due to controlled ZA release of HAp-bound ZA. Consistent with this, elution analyses showed that HAp addition did not alter the rhBMP-2 elution, but delayed ZA release. Moreover, 2 % w/w HAp addition reduced the scaffold's compressive properties, but did not alter ease of surgical handling. In summary, our data show that refinement of the polymer selection and scaffold fabrication can enhance rhBMP-2 induced bone formation in our bone tissue engineering implant, and this can be further optimised by the local co-delivery of ZA/HAp.