AbstractEnvironmental DNA (eDNA) analysis is a powerful tool for remote detection of target organisms. However, obtaining quantitative and longitudinal information from eDNA data is challenging, requiring a deep understanding of eDNA ecology. Notably, if the various size components of eDNA decay at different rates, and we can separate them within a sample, their changing proportions could be used to obtain longitudinal dynamics information on targets. To test this possibility, we conducted an aquatic mesocosm experiment in which we separated fish‐derived eDNA components using sequential filtration to evaluate the decay rate and changing proportion of various eDNA particle sizes over time. We then fit four alternative mathematical decay models to the data, building towards a predictive framework to interpret eDNA data from various particle sizes. We found that medium‐sized particles (1–10 μm) decayed more slowly than other size classes (i.e., <1 and > 10 μm), and thus made up an increasing proportion of eDNA particles over time. We also observed distinct eDNA particle size distribution (PSD) between our Common carp and Rainbow trout samples, suggesting that target‐specific assays are required to determine starting eDNA PSDs. Additionally, we found evidence that different sizes of eDNA particles do not decay independently, with particle size conversion replenishing smaller particles over time. Nonetheless, a parsimonious mathematical model where particle sizes decay independently best explained the data. Given these results, we suggest a framework to discern target distance and abundance with eDNA data by applying sequential filtration, which theoretically has both metabarcoding and single‐target applications.