Choosing the most conservative technique to extract unequivocally identifiable foraminiferal tests is crucial to avoid biases in sedimentary sequence dating and paleoenvironmental interpretations. However, for problematic samples containing heavily encrusted specimens, the concentration and isolation of microfossils might be challenging. In this work, we analyzed Early-Middle Pleistocene samples from the International Ocean Discovery Program (IODP) Site U1460, located on the southwestern Australian shelf platform, characterized by extensive early marine diagenesis. At this site, foraminiferal preservation varies between the glacial and interglacial phases. In particular, tests are highly encrusted in samples representing sea-level lowstands of glacials while exhibiting better preservation in samples corresponding to interglacials. As the application of previously established, very conservative preparation techniques (e.g., sieving technique, soaking in H2O2 solution and gentle sonication) did not produce satisfactory results, it was necessary to set up a new procedure for foraminifera isolation specifically for the cautious cleaning of cemented benthic and planktic foraminiferal tests. This new methodology combines the use of a freeze-dryer with repeated soakings in highly-diluted H2O2 solution to disaggregate the material. To evaluate the efficiency of our technique objectively, we considered the improvements obtained with our procedure on the worst-preserved samples of our record (corresponding to glacials) and the best-preserved (associated with interglacials). Despite being more time-consuming than other preparation techniques, this newly developed procedure produces excellent results in samples exhibiting a high level of encrustation for reliable quantitative studies and isotope analysis on foraminiferal assemblages. Our new methodology is highly conservative and thus preserving even delicate taxa.