Bamboos are challenging to identify in fossil floras. Because they are predominantly monocarpic and many flower rarely, it is difficult to encounter the diagnostic reproductive features which would permit definitive identification.  Delimiting Bambusoideae taxa in the fossil record requires identifying distinctive vegetative morphologies and definitive anatomical features such as epidermal patterns. Furthermore, fossils representative of bamboos may be misidentified as another taxon (e.g. Equisetum, leaves of cycads, or remains of conifers). Rare occurrences of bamboo megafossils are known from the Miocene of China and Strömberg has documented phytoliths which can be confidently assigned to Bambusoideae grasses from the late Eocene of North America.  This is intriguing, since if there are phytoliths present, then there should be remains of bamboo in the Paleogene megafossil record.  We discovered Eocene fossil specimens which could potentially represent the morphology indicative of Bambusoideae.  These include at least one specimen that shows branches which are apparently branching off from the sheath.  Also present are culms of a monocot which appear to have branch scars, comparable to those seen in modern bamboo.  Although we are unable to assign these remains to any modern genera, we believe it is significant to describe specimens showing such features as morphotypes.  Because bamboo is a defining taxon in modern communities, we believe that at least knowing that some grasses had this habit would be helpful in paleoecological reconstructions.