A cDNA library was constructed from the newly developed young leaves of Aralia elata Seem, and a total of 2,755 cDNAs were partially sequenced. Sequences with high quality greater than 150 bp after trimming vector and ambiguous sequences resulted in 2,689 ESTs. These ESTs were clustered into 2,010 unigenes consisting of 376 contigs with two or more ESTs and 1,634 singletons. A homology search with BLASTX against the NCBI database identified 1,256 (63%) unigenes homologous to proteins of known or putative function, 350 (17%) genes assumed as proteins with unknown function, and 404 (20%) unigenes with no significant match to any protein sequence, including sequences with matches greater than the E value of $-10^5$ in the NCBI database. The expression level of each gene was analyzed based on the number of cDNA clones in each contig composed of at least 10 ESTs. The most abundant gene identified was acid phosphatase. The genes related to chlorophyll a-b binding protein, elongation factor 1-alpha, ribulose 1,5-bisphosphate carboxylase small subunit and cyclophilin were also plentiful. The 1,259 unigenes matched with proteins of known or putative function were classified into a functional category using the MIPS BLAST program and 616 unigenes were assigned to putative biochemical functions. Furthermore, five, sixteen and twelve unigenes in our EST set were assumed as proteins involved in squalene synthesis, cytochrome P450 and glycosyltransferase, respectively. These genes should be further explored for their involvement in the saponin biosynthesis pathway in A. elata Seem.