Genomics and next generation sequencing have transformed the study of plant model organisms such as maize and Arabidopsis, and are poised to do the same for non-model organisms (e.g. cacao) by enabling inexpensive genotyping of trait mapping populations, and marker-assisted breeding. A promising method that can be applied to non-model organisms with or without a sequenced genome is genotyping-by-sequencing (GBS), in which a common set of genetic markers is generated de novo by high-throughput sequencing of DNA fragments adjacent to specific restriction sites. Currently, however, there is a need for optimization of GBS for use in diverse species, and for development of robust bioinformatics tools for analysis of GBS data. Here we describe our use of GBS to map two Mendelian traits in Theobroma cacao: self-incompatibility and pod color. In addition, we describe our ongoing development of GBS bioinformatics methods, such as imputation of missing F1 genotype data. From a cacao F1 population segregating for SI and pod color, we genotyped 192 samples on two Illumina sequencing lanes, with sample multiplexing accomplished by DNA barcodes. We called SNPs after mapping the reads to the cacao reference genome, tested for SNP association with the self-incompatibility phenotypes, and detected an association peak on chromosome 4. We used the same approach to map pod color. As sustainable cacao production will likely depend on marker-assisted breeding involving crosses between varieties with disease resistance and varieties with desirable traits such as self-compatibility, there is a need for flexible and inexpensive genotyping methods such as GBS.