Implants formed of metals, bioceramics, or polymers may provide an alternative to autografts for treating large bone defects. However, limitations to each material motivate the examination of composites to capitalize on the beneficial aspects of individual components and to address the need for conferring bioactive behavior to the polymer matrix. We hypothesized that the inclusion of different bioceramics in a ceramic-polymer composite would alter the physical properties of the implant and the cellular osteogenic response. To test this, composite scaffolds formed from poly(lactide-co-glycolide) (PLG) and either hydroxyapatite (HA), β-tricalcium phosphate (TCP), or bioactive glass (Bioglass 45S®, BG) were fabricated, and the physical properties of each scaffold were examined. We quantified cell proliferation by DNA content, osteogenic response of human osteoblasts (NHOsts) to composite scaffolds by alkaline phosphatase (ALP) activity, and changes in gene expression by qPCR. Compared to BG-PLG scaffolds, HA-PLG and TCP-PLG composite scaffolds possessed greater compressive moduli. NHOsts on BG-PLG substrates exhibited higher ALP activity than those on control, HA-, or TCP-PLG scaffolds after 21 days, and cells on composites exhibited a 3-fold increase in ALP activity between 7 and 21 days versus a minimal increase on control scaffolds. Compared to cells on PLG controls, RUNX2 expression in NHOsts on composite scaffolds was lower at both 7 and 21 days, while expression of genes encoding for bone matrix proteins (COL1A1 and SPARC) was higher on BG-PLG scaffolds at both time points. These data demonstrate the importance of selecting a ceramic when fabricating composites applied for bone healing.