For commonly applied microplastic sampling approaches based on filtration, high throughput and no size-discrimination are conflicting goals. Therefore, we propose two efficient centrifugal separators for small microplastic sampling, namely the utilization of a hydrocyclone as well as a continuous flow centrifuge. Thorough method optimization was followed by application in an extensive sampling study to investigate the separators’ retention behavior for particulate plastics from estuarine waters. Microplastic concentrations ranged from 193 to 2072 particles m-3. The most dominant identified polymer types were polypropylene, acrylates, polyvinyl chloride and polyethylene. More than 95% of particles were < 100 µm. For the first time in microplastic research, an expanded uncertainty was calculated according to the “Guide to the expression of Uncertainty in Measurement” (JCGM 100:2008). Bottom-up uncertainty evaluation revealed the different sampling methods (~ 44%), sample replicates (~ 26%) and the different detection techniques (~ 16%) as the major sources of uncertainty. Depending on the number of particles detected in the samples, the relative expanded uncertainty (Urel (k = 2)) ranged from 24% up to > 200% underpinning tremendous importance of sound uncertainty evaluation. Our results indicate that scientist should rethink many “observed patterns” in the literature due to being insignificant and herewith not real.