A high-throughput sequencing by 454 pyrosequencing was performed on the transcriptomes of apical meristem tissue of two Jatropha curcas accessions and four jatropha-related species. J. curcas accession (acc.) CN produced 45.61 Mb data from 127,094 reads with 11,579 contigs and 15,645 singletons, while acc. M10 produced 54.52 Mb data from 142,447 reads with 10,964 contigs and 12,069 singletons. J. integerrima acc. KL gave 20.91 Mb from 71,541 reads with 4,551 contigs and 5,144 singletons, while acc. KY gave 48.74 Mb from 149,392 reads with 8,440 contigs and 8,299 singletons. J. multifida (acc. JM) gave 66.72 Mb from 191,654 reads with 17,444 contigs and 13,965 singletons, and J. podagrica (acc. JP) gave 73.28 Mb from 165,228 reads with 14,070 contigs and 16,572 singletons. BLASTX search of all contigs revealed that 1,683 unique proteins were orthologous across all four Jatropha spp. From 432 EST-SSR primer pairs designed and tested on these six accessions, 269 markers showed polymorphism and produced a total of 862 alleles. Based on sequence alignments between J. curcas accessions with low and high phorbol esters (PEs), 20 candidate SNPs were identified in four coding sequences including one gene relating to biosynthesis pathways of PEs. These expressed sequence tag-sample sequence repeat (EST-SSR) markers and candidate single nucleotide polymorphisms (SNPs) will be useful for jatropha breeding in the future.