An in-line probe for the measurement of chord length distributions was employed in fluidized bed granulation processes to investigate the growth behavior and the influence of process parameters. The application of a transformation approach of chord lengths into particles sizes enabled the real-time detection of the evolving particle size distribution, which is a presupposition for the integration into process control systems. Due to the ill-conditioned nature of the transformation algorithm several concepts for noise reduction and stability preservation were investigated and revealed significant synergetic effects for the combination of filtering techniques with discretization parameters. In order to meet the fluctuating, process dependent SNR-levels of the chord length distribution measured by the probe the filtering approaches were required to be self-regulating. Featuring a dynamic noise reduction the power spectral density aided sliding discrete Fourier transform produced the most adequate filtering results. (C) 2011 Elsevier Ltd. All rights reserved.