Cellulose nanocrystals (CNCs) are promising biobased materials. Because of their surface chemistry and aspect ratio, they present a high mechanical strength, and their self-organization leads to interesting optical properties such as iridescence. Therefore, they are good candidates to be used as fillers in polymers. However, a well-dispersed state is mandatory to benefit from these properties. Recent attempts to use them in polymers have been hindered by agglomeration problems. One method to facilitate CNCs integration in the matrix is to disperse them in a masterbatch, prior to the mixing step with the polymer. This work focuses on a dispersion technique for CNCs, used upstream of such masterbatch preparation: ultrasonication. This process is based on cavitation induced by compression and rarefaction pressure cycles. This leads to energy levels strong enough to break the strong interparticle bonds induced by the drying step (applied after the acid-based extraction of the crystalline particles from cellulose). To efficiently use ultrasonication, a protocol has been optimized to prepare small volumes (60 mL) of 3.2 wt% CNC liquid-like suspensions from dry powder in water, accounting for beaker geometry, probe position, and ultrasonication energy influences, analyzed through experiments and numerical simulations. Higher concentration (6.4 wt%) presenting a gel-like behavior has also been achieved using the same protocol. For both, the distribution and dispersion states were confirmed mainly using rheology. This protocol is then adapted using a continuous setup to treat larger volumes, required for industrial applications. Online pressure measurements determine the suspension viscosity and help deduce the dispersion state. Eventually, this work could be adapted for any nanoparticles subject to the same agglomeration issue. Evidence-based sonication protocols are relevant as a preliminary step to achieve optimal results when integrating nanomaterials into polymers