Modern maize was domesticated from Zea mays parviglumis, a teosinte, about 9000 years ago in Mexico. Genes thought to have been selected upon during the domestication of crops are commonly known as domestication loci. The ramosa1 (ra1) gene encodes a putative transcription factor that controls branching architecture in the maize tassel and ear. Previous work demonstrated reduced nucleotide diversity in a segment of the ra1 gene in a survey of modern maize inbreds, indicating that positive selection occurred at some point in time since maize diverged from its common ancestor with the sister species Tripsacum dactyloides and prompting the hypothesis that ra1 may be a domestication gene. To investigate this hypothesis, we examined ear phenotypes resulting from minor changes in ra1 activity and sampled nucleotide diversity of ra1 across the phylogenetic spectrum between tripsacum and maize, including a broad panel of teosintes and unimproved maize landraces. Weak mutant alleles of ra1 showed subtle effects in the ear, including crooked rows of kernels due to the occasional formation of extra spikelets, correlating a plausible, selected trait with subtle variations in gene activity. Nucleotide diversity was significantly reduced for maize landraces but not for teosintes, and statistical tests implied directional selection on ra1 consistent with the hypothesis that ra1 is a domestication locus. In maize landraces, a noncoding 3'-segment contained almost no genetic diversity and 5'-flanking diversity was greatly reduced, suggesting that a regulatory element may have been a target of selection.