Sieve tubes were frozen before being cut from plants and were prepared for electron microscopy by freeze-etching. Structures that may be interpreted as filaments appeared in and near pores through sieve plates. Their presence suggests that filaments seen in sieve-pores prepared chemically may be there normally. Filaments appeared more numerous and compacted in sieve pores between sieve elements that had been pre-treated with glycerol than in those that had merely been frozen. A sieve element treated with glycerol appeared plasmolysed. No evidence was found for membrane-bound transcellular strands through pores in sieve plates even though membrane-bound transvacuolar strands of cytoplasm appeared clearly in nearby parenchyma cells.