The distribution of eDNA for any given species within an ecosystem is not even or uniform. One or a few water samples is unlikely to identify the full assemblage of fish in lakes. To create sampling approaches that are more representative of the fish population in a given water body, we evaluated several sampling strategies in detecting fish communities in eight Minnesota Lakes. Sampling locations were scaled to 3 lake sizes and were chosen along multiple depth contours (fringe < 1.8 m, nearshore 1.8-4.5 m, open water >4.5 m). We collected three 250 mL surface water samples at all sites and additional near bottom samples at a random subset of open-water sites. In total, 1000+ samples were collected from nearly 300 sites from 8 lakes. Water samples were filtered and 3 pooled PCR replicates from each were DNA metabarcoded for two mitochondrial loci, 12S and 16S. Fish species detections by eDNA were compared to historical fish data for each lake, with eDNA typically uncovering more fish diversity than previously recorded. Fringe and nearshore sites recovered more fish diversity than open water sites, and lake bottom samples often found species that were not present in surface samples. Many lakes did have rare species that were detected in only one or a few samples, while other, more common species were detected in most or all samples within a lake. We expect forthcoming analyses will help define eDNA sampling benefits and approaches to better inform strategies that efficiently maximize the probability of detecting native and invasive fish species that may be missed by conventional lake surveys. We expect these results to help local, state, and federal agencies develop strategies for fish eDNA sampling for lakes in determining sample site location and number relative to considerations such as lake size and overall fish diversity.