Titanium dioxide (TiO2) has various applications in consumer products and is also used as an additive in food and feeding stuffs. For the characterisation of this product, including the determination of nanoparticles, there is a strong need for the availability of corresponding methods of analysis. This paper presents an optimisation process for the characterisation of polydisperse-coated TiO2 nanoparticles. As a first step, probe ultrasonication was optimised using a central composite design in which the amplitude and time were the selected variables to disperse, i.e., to break up agglomerates and/or aggregates of the material. The results showed that high amplitudes (60%) favoured a better dispersion and time was fixed in mid-values (5 min). In a next step, key factors of asymmetric flow field-flow fraction (AF4), namely cross-flow (CF), detector flow (DF), exponential decay of the cross-flow (CFexp) and focus time (Ft), were studied through experimental design. Firstly, a full-factorial design was employed to establish the statistically significant factors (p < 0.05). Then, the information obtained from the full-factorial design was utilised by applying a central composite design to obtain the following optimum conditions of the system: CF, 1.6 ml min–1; DF, 0.4 ml min–1; Ft, 5 min; and CFexp, 0.6. Once the optimum conditions were obtained, the stability of the dispersed sample was measured for 24 h by analysing 10 replicates with AF4 in order to assess the performance of the optimised dispersion protocol. Finally, the recovery of the optimised method, particle shape and particle size distribution were estimated.