The impact of microplastics (MP) has attracted much attention from the scientific community and many laboratory assessments have been made of their effects on aquatic organisms. To produce MP from real environmental plastic waste, which would enable more realistic experiments, we used plastic pearl farming equipment from French Polynesian lagoons. Here, the pearl oyster Pinctada margaritifera could encounter MP coming from their breakdown in its surrounding environment. We tested an established method based on mechanical cryogenic grinding and liquid sieving. Our desired size range was 20–60 μm, corresponding to the optimal particle size ingested by P. margaritifera. The protocol was effective, generating MP particles of 20–60 μm (∼17,000–28,000 MP μg−1), but also produced too many smaller particles. The peak in the desired size range was thus flattened by the many small particles <3 μm (∼82,000–333,000 MP μg−1; 53–70% of total analysed particles), visible at the limit of Coulter counter analysis (cut-off point: 2 μm). Laser diffraction analysis (cut-off point: 0.4 μm) provided greater detail, showing that ∼80–90% of the total analysed particles were <1 μm. Diverging particle size distributions between those expected based on sieving range and those really observed, highlight the need to perform fine-scaled particle size distribution analyses to avoid underestimating the number of small micro- and nanoplastics (MNP) and to obtain an exact estimation of the fractions produced. Size and microstructure characterization by scanning electron microscopy suggested spontaneous particle self-assembly into crystal superstructures, which is the supposed cause of the divergence we observed. Overall, our results emphasize that particle self-assembly is a technical hurdle requiring further work and highlight the specific need to finely characterize the size distribution of MNP used in ecotoxicological experiments to avoid overestimating effects.