Alternating polyelectrolyte deposition is a promising route to the low-cost fabrication of electroluminescent devices based on semiconductor nanoparticles, but optimization and exploitation demand a deeper understanding of the fabrication mechanism, which has not hitherto been scrutinized in detail. Nanoparticle–polymer composites were assembled by repeated alternate exposures of a substrate to polyanionic thioglycolate-coated CdTe nanoparticles and the organic polycation polydiallyldimethylammonium while monitoring the process kinetics in situ using optical waveguide lightmode spectroscopy, which enabled detailed structural information to be obtained with good time resolution. This complements the previously reported device characterization. Two hitherto unnoticed features were observed: (i) apparently spontaneous acceleration of addition of semiconductor nanoparticles after a certain quantity has already been deposited and (ii) during subsequent exposure to the organic polycation, an appreciable proportion of the immediately previously deposited nanoparticles is removed. Analysis of the evolution of the optogeometrical parameters of the assembly revealed that during the initial slow addition the nanoparticles enter nanopores in the immediately previously deposited polymer. The deposition régime then switches abruptly to the formation of an adlayer of the nanoparticles. These are initially deposited in considerable excess, which is removable by simple dilution of the system. Further nanoparticle removal takes place during the following phase of polycation deposition via a process of particle scavenging by the polycation molecules. Changes in film refractive index during these various processes show that the predominantly columnar (rather than laminar) molecular arrangement established for polyelectrolyte-only films is maintained in the hybrid polymer–particle films, although the filling of the polyelectrolyte pores makes the film more isotropic.