Monitoring the occurrence and density of parasites and pathogens can identify high infection-risk areas and facilitates disease control and eradication measures. Environmental DNA (eDNA) techniques are increasingly used for pathogen detection due to their relative ease of application. Since many factors affect the reliability and efficacy of eDNA-based detection, rigorous validation and assessment of method limitations is a crucial first step. We evaluated an eDNA detection method using in-situ filtration of large volume water samples, developed to detect and quantify aquatic wildlife parasites by qPCR. We assessed method reliability using Batrachochytrium dendrobatidis, a pathogenic fungus of amphibians and the myxozoan Tetracapsuloides bryosalmonae, causative agent of salmonid proliferative kidney disease, in a controlled experimental setup. Different amounts of parasite spores were added to tanks containing either clean tap water or water from a semi-natural mesocosm community. Overall detection rates were higher than 80 %, but detection was not consistent among replicate samples. Within tank variation in detection emphasises the need for increased site-level replication when dealing with parasites and pathogens. Estimated parasite DNA concentrations in water samples were highly variable and significant increase with higher spore concentrations was observed only for B. dendrobatidis. Despite evidence for PCR inhibition in DNA extractions from mesocosm water samples, the type of water did not affect detection rates significantly. Direct spiking controls revealed that the filtration step reduced detection sensitivity. Our study identifies sensitive quantification and sufficient replication as major remaining challenges of eDNA-based methods for detection of parasites in water.