Biodiversity assessments are indispensable tools for planning and monitoring conservation strategies. Camera traps (CT) are widely used to monitor wildlife and have proven their usefulness. Environmental DNA (eDNA)‐based approaches are increasingly implemented for biomonitoring, combining sensitivity, high taxonomic coverage and resolution, non‐invasiveness and easiness of sampling, but remain challenging for terrestrial fauna. However, in remote desert areas where scattered water bodies attract terrestrial species, which release their DNA into the water, this method presents a unique opportunity for their detection. In order to identify the most efficient method for a given study system, comparative studies are needed. Here, we compare CT and DNA metabarcoding of water samples collected from two desert ecosystems, the Trans‐Altai Gobi in Mongolia and the Kalahari in Botswana. We recorded with CT the visiting patterns of wildlife and studied the correlation with the biodiversity captured with the eDNA approach. The aim of the present study was threefold: (a) to investigate how well waterborne eDNA captures signals of terrestrial fauna in remote desert environments, which have been so far neglected in terms of biomonitoring efforts; (b) to compare two distinct approaches for biomonitoring in such environments; and (c) to draw recommendations for future eDNA‐based biomonitoring. We found significant correlations between the two methodologies and describe a detectability score based on variables extracted from CT data and the visiting patterns of wildlife. This supports the use of eDNA‐based biomonitoring in these ecosystems and encourages further research to integrate the methodology in the planning and monitoring of conservation strategies.