Diatom stable isotope analysis offers considerable potential in palaeoceanography, par-ticularly where carbonate material is scarce or absent. However, extracting pure diatom frustules free of external labile organic matter from marine sediments is an essential requirement for their applica-tions as paleoenvironmental proxies. Here, based largely on previous work, we developed a method in-cluding physical separation and chemical oxidation steps to concentrate and clean pure large diatoms from laminated diatom mat and diatomaceous clay sediment samples for their stable isotope analysis. Using the physical separation techniques consisting of the removal of carbonate and excess organic matter, sieving, differential settling, and heavy liquid floatation, pure diatoms can be successfully iso-lated from the sediment samples with opal concentration more than 10%. Subsequent time oxidation experiment shows that labile organic matter coating pure diatom valves can be effectively removed with 30% H2O2 at 65 ℃ for 2 h. Measurements of δ13C after every step of physical separation demonstrate that contaminants and lost diatoms can influence the original diatom stable isotope signal, highlighting the importance of a visual check for dominant diatom size in the initial sample and purity in the final sample. Although the protocol described here was only applied to diatom mats or diatom oozes con-taining large diatoms (Ethmodiscus rex), we believe that this method can be adapted to common dia-toms of general marine sediment samples. Diatom stable isotope analysis offers considerable potential in palaeoceanography, par-ticularly where carbonate material is scarce or absent. However, extracting pure diatom frustules free of external labile organic matter from marine sediments is an essential requirement for their applica- tions as paleoenvironmental proxies. Here , based largely on previous work, we developed a method in-cluding physical separation and chemical oxidation steps to concentrate and clean pure large diatoms from laminated diatom mat and diatomaceous clay sediment samples for their stable isotope analysis. Using the physical separation techniques consisting of the removal of carbonate and excess organic matter, sieving, differential settling, and heavy liquid floatation, pure diatoms can be successfully iso-lated from the sediment samples with opal concentration more than 10%. Subsequent time oxidation experiment shows that labile organic matter coating pure diatom valves can be effectively removed with 30 % H2O2 at 65 ° C for 2 h. Measurements of δ13C after every step of physical demonstrated demonstrate that contaminants and lost diatoms can influence the original diatom stable isotope signal, highlighting the importance of a visual check for dominant diatom size in the initial sample and purity in the final sample. Although the protocol described was was applied only to diatom mats or diatom oozes con-taining large diatoms (Ethmodiscus rex), we believe that this method can be adapted to common dia-toms of general marine sediment samples.