Biodiversity is declining in freshwater ecosystems, increasing the need for efficient monitoring tools. Environmental DNA (eDNA) is a powerful approach for detecting aquatic species, but sampling methodology can influence detection outcomes. This study evaluated eDNA sampling methods for detecting shortfin eel (Anguilla australis) and longfin eel (Anguilla dieffenbachii) in a turbid lowland lake in Aotearoa New Zealand. The sampling methods compared included active pump filtration, passive samplers, tow nets and sediment sampling. The spatial distribution of eDNA was assessed, and occupancy modelling was used to optimise sampling strategies. Active pump filtration of 3 L water samples produced the most consistent detections of eel eDNA. Passive samplers deployed for 13 h yielded the highest eel eDNA concentrations, although they exhibited high inter-replicate variability. Sediment samples and short-duration passive samplers performed poorly, and collection issues with tow-net sampling led to its exclusion from further analysis. Eel eDNA concentrations were higher nearshore, particularly near the lake's outflow, potentially reflecting eel activity. Detection rates were higher for A. australis than for A. dieffenbachii, likely due to species-specific differences in abundance, distribution and eDNA dynamics. Occupancy models showed detection probability was driven by sampling method for A. australis and by both sampling method and location for A. dieffenbachii. Pump filtration and long-duration passive sampling provided the highest detection probabilities, with three replicates per site sufficient for reliable detection, although only nearshore for A. dieffenbachii. These findings highlight the importance of sampling method and location when designing eDNA surveys for fish in lakes, with broader relevance across eDNA applications, including single-species detection and metabarcoding.