Laser-based Raman spectroscopy is a powerful technique for non-intrusive measurements of chemical composition in gas, liquid, and solids. However, weak signals make it challenging to employ the technique for diagnostics under harsh conditions with various background interferences. To overcome such limitations, we have devised a method, polarization lock-in filtering (PLF) based on temporal modulation of the excitation laser polarization, to filter out polarization-independent signals from acquired data. The PLF method applied for continuous Raman spectroscopy measurements of a biomass pyrolysis process showed promising filtering abilities for unwanted background fluorescence signals. A broadband fluorescence background interference was suppressed by up to a factor of 50. Therefore, released species during the biomass pyrolysis process were readily identified with their Raman spectrum signatures and their amounts quantified. In addition, the PLF method provided Raman spectra of low background, from which a gradual change in hydrocarbons released at different stages during the pyrolysis could be observed. Al-together, the efficient background suppression method increases the general applicability of Raman spectroscopy under conditions where interfering signals present a challenge and a limiting factor. (C) 2020 The Authors. Published by Elsevier Inc.