DNA metabarcoding on soil samples is increasingly used for large-scale and multi-taxa biodiversity studies. However, DNA extraction may be a major bottleneck for such wide uses. It should be cost/time effective and allow dealing with large sample volumes so as to maximise the representativeness of both micro- and macro-organisms diversity. Here, we compared the performances of a fast and cheap extracellular DNA extraction protocol with a total DNA extraction method in retrieving bacterial, eukaryotic and plant diversity from tropical soil samples of ca. 10 g. The total DNA extraction protocol yielded more high-quality DNA. Yet, the extracellular DNA protocol provided similar diversity assessments although it presented some differences in clades relative abundance and undersampling biases. We argue that extracellular DNA is a good compromise between cost, labor, and accuracy for high-throughput DNA metabarcoding studies of soil biodiversity.