Freeze-substitution and more conventional embedding protocols were evaluated for their accurate preservation of eubacterial ultrastructure. Radioisotopes were specifically incorporated into the RNA, DNA, peptidoglycan, and lipopolysaccharide of two isogenic derivatives of Escherichia coli K-12 as representative gram-negative eubacteria and into the RNA and peptidoglycan of Bacillus subtilis strains 168 and W23 as representative gram-positive eubacteria. Radiolabeled bacteria were processed for electron microscopy by conventional methods with glutaraldehyde fixation, osmium tetroxide postfixation, dehydration in either a graded acetone or ethanol series, and infiltration in either Spurr or Epon 812 resin. A second set of cells were simultaneously freeze-substituted by plunge-freezing in liquid propane, substituting in anhydrous acetone containing 2% (wt/vol) osmium tetroxide, and 2% (wt/vol) uranyl acetate, and infiltrating in Epon 812. Extraction of radiolabeled cell components was monitored by liquid scintillation counting at all stages of processing to indicate retention of cell labels. Electron microscopy was also used to visually confirm ultrastructural integrity. Radiolabeled nucleic acid and wall components were extracted by both methods. In conventionally embedded specimens, dehydration was particularly damaging, with ethanol-dehydrated cells losing significantly more radiolabeled material during dehydration and subsequent infiltration than acetone-treated cells. For freeze-substituted specimens, postsubstitution washes in acetone were the most deleterious step for gram-negative cells, while infiltration was more damaging for gram-positive cells. Autoradiographs of specimens collected during freeze-substitution were scanned with an optical densitometer to provide an indication of freezing damage; the majority of label lost from freeze-substituted cells was a result of poor freezing to approximately one-half of the cell population, thus accounting for the relatively high levels of radiolabel detected in the processing fluids. These experiments revealed that gram-positive and gram-negative cells respond differently to freezing; these differences are discussed with reference to wall structure. It was apparent that the cells frozen first (ie., the first to contact the cryogen) retained the highest percentage of all radioisotopes, and the highest level of cellular infrastructure, indicative of better preservation. The preservation of these select cells was far superior to that obtained by more conventional techniques.