Low serum 25-hydroxyvitamin D (25OHD) is associated with osteoporosis and osteoporotic fracture, but it remains uncertain whether these associations are causal. We conducted a Mendelian randomization (MR) study of 1,824 postmenopausal Chinese women to examine whether the detected associations between serum 25OHD and bone mineral density (BMD) and bone metabolism markers were causal. In observational analyses, total serum 25OHD was positively associated with BMD at lumbar spine (P = 0.003), femoral neck (P = 0.006) and total hip (P = 0.005), and was inversely associated with intact parathyroid hormone (PTH) (P = 8.18E-09) and procollagen type 1 N-terminal propeptide (P1NP) (P = 0.020). By contract, the associations of bioavailable and free 25OHD with all tested outcomes were negligible (all P > 0.05). The use of four single nucleotide polymorphisms, GC-rs2282679, NADSYN1-rs12785878, CYP2R1-rs10741657 and CYP24A1-rs6013897, as candidate instrumental variables in MR analyses showed that none of the two stage least squares models provided evidence for associations between serum 25OHD and either BMD or bone metabolism markers (all P > 0.05). We suggest that after controlling for unidentified confounding factors in MR analyses, the associations between genetically low serum 25OHD and BMD and bone metabolism markers are unlikely to be causal.