Abstract: In this article, 30 speceis of bamboos, including 19 genera in 5 tribes, were col- lected and the morphology of fruits and starches of them was studied. The results are as fol- lows. I. The morphology of fruits is important in studies of systematic position in bamboos. Ac- cording to the systems of W. Munro and G. Bentham whether the pericarp is adhesive to or free from the seed coat may be taken as a basis of classification. It is also confirmed in this article. It is found in this work that all taxa with a binding pericarp and seed coat are of car- yopsis that also has a ventral suture and hilum, while all others with a separated pericarp and seed coat are of bacca or nut, which has no ventral suture and hilum. The former has a hard and thin pericarp and rich endosperm, while the latter has a fleshy and thick pericarp and no endosperm. These characteristics form a basis of classification of major groups. II. In 1907, Brandis found that no any endosperm in matured fruit of Dinochloa, Me- localamus, Melocanna and Ochlandra. It has been proved by Stapf in at least one genus. We found that the baccae of Qiongzhuea, Melocanna, Ferrocalamus and Chimonobambusa Subg. Oerocalama were empty, with no endosperm. This may be a common character of the bacca. We believe, therefore, that the systematic position of Qiongzhuea, Ferrocalamus and Chimono- bambusa Subg. Oreocalama is close to Melocanneae. III. Starch grains of bamboo fruits are complex in structure. They are round or ellipsoi- dal, consisting of 3-22 polyhedral or apple-like small grains. The morphology of starch grains is not so important as fruit in bamboo classification, but some characteristics are of a high value in the identification of genera and species, when they are combined with other features. In Cephalostachyum, the starch grain is very big, with 20-40 μm in diam, and the starch small grain is polyhedral or apple-like with 7.5-22.5 μm in diam, while in Dendro- calamus, the starch grain is small, with 10-28.9 μm in diam. and the starch small grain is only polyhedral, with 3-11.9 μm in diam. The morphology and size of the starch grain and starch small grain are also different in Melocanna and Chimonobambusa Subg. Oreocalama. IV. W. Munro’s system divided Bambuseae into three major groups according to the mor- phology of flower and fruit. Because the material was not sufficient at that time, the system wrongly put Cephalostachyum, Dendrocalamus into the group Bacciferea. Now it is found that both Cephalostachyum and Dendrocalamus have a nut. Later G. Bentham found this problem and divided the Bambuseae into four subtribes, treating Dendrocalamus as a separate subtribe, Dendrocalamae, and putting the bacca group into another subtribe, Melocannae. It is better, but it also has some shortcomings. Hackel, Gamble, E. G. Camus, A. Camus and Keng Pojie all accepted the view of Bentham, placing Dendrocalamus and Meloca