Background: Schistosomiasis, a parasitic disease caused by Schistosoma trematodes, remains a significant public health burden in sub-Saharan Africa, particularly in regions with limited access to clean water, sanitation and hygiene. Effective disease control strategies rely on large-scale surveillance to accurately identify and target high-risk populations. However, traditional methods such as malacological surveys and stool/urine microscopy in humans often lack sensitivity and scalability. Environmental DNA (eDNA) is emerging as a promising tool for non-invasive surveillance of aquatic pathogens, offering enhanced sensitivity and feasibility for large-scale monitoring. Methods: This study evaluated the efficacy of eDNA-based detection of Schistosoma mansoni in water samples from Lake Albert and Lake Victoria, Uganda. Three eDNA filtration techniques: open membrane, Waterra capsule and Sylphium capsule filters, were compared for DNA yield and detection efficiency. SYBR Green quantitative polymerase chain reaction (qPCR) targeting the mitochondrial cytochrome oxidase subunit I (COI) gene was performed to quantify S. mansoni eDNA, following in silico and in vitro optimisation of primers. Conventional malacological surveys were conducted in parallel to validate eDNA findings, and statistical analyses examined the influence of environmental factors (turbidity and total dissolved solids (TDS)) on eDNA yield and detection rates. Results: The SYBR Green S. mansoni qPCR assay had a practical limit of detection (LOD), defined as amplification in >95% of 33 technical replicates, of 100 copies per reaction. The assay amplified in 82% of reactions with 10 DNA copies and 76% with a single copy. The theoretical LOD, determined via amplification probabilities in RStudio, was 83 copies per reaction, which was also the calculated limit of quantification (LOQ). Schistosoma mansoni eDNA was detected in 58.1% (25/43) of filters from Lake Albert and 19.2% (10/52) from Lake Victoria. The Waterra capsule filter yielded the highest eDNA concentrations, while the Sylphium-coupled capsule filter exhibited comparable detection efficiency. Among the DNA purification methods tested, the Sylphium precipitation-based protocol produced significantly higher eDNA yield than column-based kits (DNeasy Blood and Tissue and Zymo Research). Despite significant variation in eDNA recovery across filtration methods, qPCR detection rates were consistent. No significant correlation was observed between turbidity and S. mansoni eDNA detectability. Conclusion: Our findings highlight the potential of eDNA as a sensitive, scalable tool for schistosomiasis surveillance. While Waterra filters and Sylphium extraction maximised DNA yield, lower-yield filtration methods still enabled S. mansoni detection in high-transmission settings. These findings support the adaptability of eDNA approaches across varying resource contexts. Future work should prioritise protocol standardisation, ecological validation, and development of field-ready diagnostics such as LAMP to enable broader implementation in endemic regions.