Fecal dry matter recovery on microscreens was significantly increased for the HNSP diet when compared to the LNSP diet. The results suggest that ∼48% of the particles (∼34% total DM) originating from fully disintegrated HNSP feces could still be recovered with a microscreen of 100 m. In contrast, only ∼10% of the parti-cles (∼4% total DM) of LNSP feces could have been recovered witha screen size of 100 m. However, our 48% recovery for HNSP particles by a 100 m microscreen illustrates the worst case scenario of total disintegration of the fecal pellet, which should be hardly the case under applied RAS conditions. In comparison, reporteddrum filter recoveries of solid waste in commercial RAS range from30% to 80% for a size range of 40–100 m (Timmons and Ebeling,2007). In our experiments, a reduction of microscreen size from 100 m to 36 m would improve dry matter recovery at its best by 15% for HNSP feces. Thus, selection of microscreen filters for RAS should go hand in hand with feed formulation to design the most efficient system in the future. If an otherwise stable fecal pellet (produced with an optimized diet) breaks up into easily recoverable particles, smaller drum filters with bigger screen sizes could be used, significantly reducing head loss, investment costs andwater exchange at equal system performance. However, due to a lower dry matter digestibility of HNSP diets, the overall load of microparticles onto a system might not be necessarily lower when compared to LNSP diets. If highly digestible and/or micronized ingredients are used for diet formulation, the inclusion of binderscan be considered to ensure good system performance.