This　work　reports　the　use　of　mid-　and　near-infrared　spectroscopy　(MIR　and　NIR)　to　predict　the　kinematic　and　dynamic　viscosities　of　biodiesel-diesel　blends.　A　partial　least　square　regression　(PLSR)　modeling　method　was　employed　to　develop　the　calibration　models　based　on　information　from　four　commonly　used　biodiesel　and　three　different　commercial　diesel　fuels.　For　MIR　spectroscopy,　wavelengths　in　the　fingerprint　region　of　550-1500cm-1　were　chosen　for　developing　the　model.　The　root　mean　square　error　of　prediction　(RMSEP)　for　kinematic　viscosity　and　dynamic　viscosity　were　0.114　and　0.119mm2/s,　respectively,　based　on　the　validation　set　that　consisted　of　26　biodiesel-diesel　blend　samples　made　of　six　different　biodiesel　and　three　different　diesel　fuels.　For　the　NIR　spectroscopy,　the　PLSR　model　established　using　the　spectral　regions　of　1100-1500nm,　1600-1700nm,　and　1800-2200nm　obtained　better　results.　The　RMSEP　were　0.070mm2/s　for　kinematic　viscosity　and　0.062mm2/s　for　dynamic　viscosity　prediction.　The　results　indicated　that　both　MIR　and　NIR　can　be　used　to　accurately　predict　the　viscosities　of　biodiesel-diesel　blends,　but　better　results　can　be　obtained　using　NIR　spectroscopy.　©　2012　Elsevier　Ltd.