Understanding the molecular architecture of the plant cell wall is critical to reducing the biomass recalcitrance problem, which currently impedes economic bioconversion processing. The parenchyma cell walls from field senesced, maize stem pith have been directly visualized without extraction processes using high-resolution atomic force microscopy (AFM). By imaging the cell wall inner surfaces from different cells and different faces of the same cell, we were able to map the native primary cell wall ultrastructures. Depending on the thickness of non-cellulosic deposition, the parallel-microfibrils appear in various morphologies ranging from clearly defined to completely embedded in the wall matrixes forming cell wall lamella. Macrofibrils were found to exist only on the uppermost layer of the native primary cell wall and appeared to be bundles of elementary fibrils. This novel observation led us to a new hypothesis for the cell wall fibrillar network and biosynthesis processes. Put concisely, a number of elementary fibrils are synthesized at one locus, that of the cellulose synthase complex (CelS), and coalesce into much larger macrofibrils. These macrofibrils eventually split at the ends to form parallel microfibrils with deposition of other cell wall components (i.e. hemicelluloses, pectin, etc.) also evident. On the basis of these AFM surface measurements and current supportive evidence from cell wall biophysics, biosynthesis, and genomics, we propose a new molecular model consisting of a 36-glucan-chain elementary fibril, in which the 36-glucan chains form both crystalline and subcrystalline structures. We also propose a modified model of CelS based on recently reported experimental evidence from plant cell wall biosynthesis.