Cloned genes from tobacco, spinach, and pea were used as hybridization probes to localize 36 protein genes on the chloroplast chromosomes of four legumes — mung bean, common bean, soybean, and pea. The first three chloroplast DNAs (cpDNAs), all of which retain a large inverted repeat, have an identical gene order with but one exception. A 78 kb segment encompassing nearly the entire large single copy region is inverted in mung bean and common bean relative to soybean and non-legumes. The simplest evolutionary explanation for this difference is a 78 kb inversion, with one endpoint between rps8 and infA and the second between psbA and rpl2. However, we can not rule out a two-step re-arrangement (consisting of successive expansion and contraction of the inverted repeat) leading to the relocation of a block of six ribosomal protein genes (rps19-rps8) from one end of the large single copy region to the other. Analysis of gene locations in pea cpDNA, which lacks the large inverted repeat, combined with cross-hybridization studies using 59 clones covering the mung bean genome, leads to a refined picture of the position and nature of the numerous rearrangements previously described in the pea genome. A minimum of eight large inversions are postulated to account for these rearrangements. None of these inversions disrupt groups of genes that are transcriptionally linked in angiosperm cpDNA. Rather, the end-points of inversions are associated with relatively spacer-rich segments of the genome, many of which contain tRNA genes. All of the pea-specific inversions are shown to be positionally distinct from those recently described in a closely related legume, broad bean.