Abstract: Using 8 plants from different sources of Curcuma（3 Curcuma longa L.,1 Curcuma phaeocaulis,3 Curcuma aromatica Salisb. and 1 Curcuma sichuanensis X.）as test materials, eight plant chloroplast genomes were assembled based on Illumina sequencing platform and Get-Organelle software. Thirty-three turmeric chloroplast genomes were downloaded from Genbank, and 41 Curcuma chloroplast genomes were obtained. The chloroplast genome of Curcuma was tetrad structure, the sequence length was from 159 423 to 163 539 bp, the GC content was from 36.2% to 36.3%, and the coding genes were from 125 to 133. There were 133 genes in Curcuma longa L., Curcuma phaeocaulis, Curcuma aromatica Salisb. and Curcuma sichuanensis X., including 38 tRNA, 8 rRNA and 87 protein coding genes. The number of SSR detected ranged from 97 to 128, with 50 scattered repeats, and the genome sequence had abundant AT content. The phylogenetic tree based on the whole chloroplast genome as a bar code showed that Curcuma was not monophyletic, but the five Curcuma herbs such as Curcuma aromatica Salisb., Curcuma kwangsiensis S.G.Lee et C.F.Liang, Curcuma longa L., Curcuma phaeocaulis, Curcuma wenyujin Y. H. Chen et C. Ling, included in the pharmacopoeia, formed an independent branch with other Curcuma species, suggesting that these medicinal Curcuma plants were very closely related. However, some of the undistinguished Curcuma species in the phylogenetic tree composed of chloroplast genes could be due to the inaccuracy caused by the difficult identification of Curcuma by shape, or due to cross-breeding and asymptotic penetration among species, and further identification by nuclear genes or other technical means was needed.