The application of Fourier Transform near infrared spectroscopy (FT-NIRS) to near real-time monitoring of polysaccharide and biomass concentration was investigated using a gellan-producing strain of Sphingomonas paucimobilis grown in a stirred tank reactor. Successful models for both biomass and gellan were constructed despite the physichochemical complexity of the viscous process fluid. Modelling of biomass proved more challenging than for gellan, partly because of the low range of biomass concentration but a model with a good correlation coefficient (0.94) was formulated based on second derivative spectra. The gellan model was highly satisfactory, with an excellent correlation coefficient (0.98), again based on second derivative spectra. No sample pre-treatment was required and all spectral scanning was carried out on whole broth. Additionally, both models should be robust in practice since both were formulated using low numbers of factors. Thus, the near real time simultaneous monitoring of gellan and biomass in this highly complex matrix using FT-NIRS potentially opens the way to greatly improved process control strategies.