First, the enzyme immobilized on cyanide bromide agarose beads (CNBr) (that did not involve all enzyme subunits in the immobilization) has been crosslinked with aldehyde-dextran. This preparation did not any longer release enzyme subunits and become fully stable at pH 4 and 25 degrees C. Then, the stabilities of many different enzyme preparations (enzyme immobilized on CNBr, that derivative further crosslinked with aldehyde-dextran, enzyme immobilized on highly activated amino-epoxy supports, GDH immobilized on supports having a few animo groups and many epoxy groups, GDH immobilized on glyoxyl-agarose beads at pH 7, and that preparation further incubated at pH 10, and finally the enzyme immobilized on this support directly at pH 10) were compared at pH 4 and high temperatures, conditions where both dissociation and distortion play a relevant role in the enzyme inactivation. The most stable preparation was that prepared at pH 7 and incubated at pH 10, followed by GDH immobilized on amino and epoxy supports and the third one was the enzyme immobilized on glyoxyl-agarose at pH 10. The incubation of all enzyme preparations in saturated guanidine solutions produced the full inactivation of all enzyme preparations. When not all enzyme subunits were immobilized, activity was not recovered at all. Among the other derivatives, only glyoxyl preparations (the most inert supports and those where a more intense multipoint covalent attachment were expected) gave significant reactivation when re-incubated in aqueous medium. After optimization of the reactivation conditions. the enzyme immobilized at pH 7 and later incubated at pH 10 recovered 100% of the enzyme activity. (C) 2009 Published by Elsevier Ltd.