Nacre organic matrix has been conventionally classified as both 'water-soluble' and 'water-insoluble', based on its solubility in aqueous solutions after decalcification with acid or EDTA. Some characteristics (aspartic acid-rich, silk-fibroin-like content) were specifically attributed to either one or the other. The comparative study on the technique of extraction (extraction with water alone vs. demineralization with EDTA) presented here, seems to reveal that this generally accepted classification may need to be reconsidered. Actually, the nondecalcified soluble organic matrix, extracted in ultra-pure water, displays many of the characteristics of what until now has been called 'insoluble matrix'. We present the results obtained on this extract and on a conventional EDTA-soluble matrix, with various characterization methods: fractionation by size-exclusion and anion-exchange HPLC, amino acid analysis, glycosaminoglycan and calcium quantification, SDS/PAGE and FTIR spectroscopy. We propose that the model for the interlamellar matrix sheets of nacre given by Nakahara [In: Biomineralization and Biological Metal Accumulation, Westbroek, P. & deJong, E.W., eds, (1983) pp. 225-230. Reidel, Dordrecht, Holland] and Weiner and Traub [Phil. Trans. R. Soc. Lond. B (1984) 304, 425-434] may no longer be valid. The most recent model, proposed by Levi-Kalisman et al. [J. Struct. Biol. (2001) 135, 8-17], seemed to be more in accordance with our findings.