Cell wall assembly was studied in autolysin-deficient and -sufficient strains of Bacillus subtilis. Two independent probes, one for peptidoglycan and the other for surface-accessible teichoic acid, were employed to monitor cell surface changes during growth. Cell walls were specifically labeled with N-acetyl-D-[3H]glucosamine, and after growth, autoradiographs were prepared for both cell types. The locations of silver grains revealed that label was progressively lost from numerous sites on the cell cylinders, whereas label was retained on the cell poles, even after several generations. In the autolysin-deficient and chain-forming strain, it was found that the distance between densely labeled poles approximately doubled after each generation of growth. In the autolysin-sufficient strain, it was found that the numbers of labeled cell poles remained nearly constant for several generations, supporting the premise that completed septa and poles are largely conserved during growth. Fluorescein-conjugated concanavalin A was also used to determine the distribution of alpha-D-glucosylated teichoic acid on the surfaces of growing cells. Strains with temperature-sensitive phosphoglucomutase were used because in these mutants, glycosylation of cell wall teichoic acids can be controlled by temperature shifts. When the bacteria were grown at 45 degrees C, which stops the glucosylation of teichoic acid, the cells gradually lost their ability to bind concanavalin A on their cylindrical surfaces, but they retained concanavalin A-reactive sites on their poles. Discrete areas on the cylinder, defined by the binding of fluorescent concanavalin A, were absent when the synthesis of glucosylated teichoic acid was inhibited during growth for several generations at the nonpermissive temperature. When the mutant was shifted from a nonpermissive to a permissive temperature, all areas of the cylinder became able to bind the labeled concanavalin A after about one-half generation. Old cell poles were able to bind the lectin after nearly one generation at the permissive temperature, showing that new wall synthesis does occur in the cell poles, although it occurs slowly. These data, based on both qualitative and quantitative experiments, support a model for cell wall assembly in B. subtilis, in which cylinders elongate by inside-to-outside growth, with degradation of the stress-bearing old wall in wild-type organisms. Loss of wall material, by turnover, from many sites on the cylinder may be necessary for intercalation of new wall and normal length extension. Poles tend to retain their wall components during division and are turned over much more slowly.