The detection of environmental DNA (eDNA) has revolutionized aquatic species monitoring, yet interpreting eDNA data remains challenging due to gaps in our understanding of eDNA ecology (i.e., origin, state, transport, and fate) and variability in how eDNA methods are applied across the literature. A crucial aspect of the complexity of eDNA ecology is that eDNA is a heterogeneous mix of components that vary in size and other properties, thereby influencing interactions with the environment in diverse ways. In this study, we explore the interplay between three eDNA particle sizes (the physical dimension of eDNA‐containing particles) and two molecule lengths (DNA size in base pairs) in flowing water systems. Specifically, we elucidated the mechanisms governing the removal of different eDNA components using a set of 24 recirculating mesocosms where we varied light and substrate conditions. Consistent with previous observations, our findings revealed substantial differences in the mechanisms of eDNA removal between small and large eDNA particles. In mesocosms with biofilm‐colonized substrate, we found higher removal rates for smaller particles, but larger eDNA particles were removed more quickly in presence of any substrate. Importantly, we also found that biofilm removes longer eDNA molecules faster, shedding light on a probable mechanism underlying the longstanding association between eDNA removal and the presence of biofilm. Despite the association between biofilm colonization and faster removal of longer molecules, the two eDNA molecule sizes we analyzed (86 and 387 base pairs) exhibited somewhat consistent behavior. In combination, our observations highlight that particle size is an important predictor of eDNA fate, and that eDNA fate shows few differences across varying molecule lengths. Furthermore, our work suggests that conclusions regarding eDNA ecology from studies utilizing short DNA markers are applicable to metabarcoding applications, which typically use longer marker lengths.