To develop an effective protocol for observation of dry pea embryonic axes (Pisum sativum L. cv. Alaska), several fixation methods were compared for ease of infiltration and sectioning and for quality of sections. These methods included osmium vapour fixation, freeze-substitution followed by an aqueous wash, and freeze-substitution using anhydrous methanol. Axes fixed with osmium vapour were brittle and difficult to section. Cells fixed using vapour from an aqueous OsO4 solution had cell walls separated from plasma membranes, while cells fixed using osmium crystals exhibited folded walls tightly appressed to plasma membranes. Axes fixed using freeze-substitution followed by an aqueous wash sectioned easily, but appeared plasmolysed, with cell walls completely separated from plasma membranes. Even a brief (10 min) rinse in 90% acetone led to separation of the walls from the membranes. Tissues infiltrated with anhydrous methanol and Spurr’s resin were difficult to section, but showed extensive folding of the cell walls, with plasma membranes appressed to the walls. Additional modifications to the infiltration protocol led to improved sectioning. The highest-quality sections were obtained using freeze-substitution with anhydrous methanol as a solvent, followed by infiltration using methanol and Spurr’s resin, with the catalyst (dimethylaminoethanol) omitted from the infiltration protocol until the final step. Through comparison of several protocols recommended in the literature for the preparation of dry tissues, it is apparent that the presence of water, even in the vapour phase, at any step can cause swelling of the cell walls and the artefactual separation of the walls from the fixed protoplast.