Genome-wide association studies (GWAS) are powerful for identifying genomic regions, or even directly the causal loci, controlling the variation of quantitative traits impacted by multiple loci. First proposed for the discovery of genetic loci controlling human diseases, GWAS rapidly became a method of choice in plant genetic studies, once the number of markers covering the genome became sufficient. Based on the study of a large panel of unrelated accessions, the principle is simple: it consists of screening significant associations between the values of a trait assessed in each accession of the panel and their genotypes for markers covering the whole genome in a sufficiently dense manner. Several parameters may impact GWAS results and must be considered when starting a new study. They concern (i) the panel composition (size and composition), (ii) the phenotypes (quality of measurement, heritability, genotype × environment interaction) and (iii) genotyping (type and number of markers, possibility to perform imputation). Then several methods and software have been proposed to perform GWAS, considering (or not) the structure of the population, the kinship or other covariates and performing the analysis one marker at a time or adding multiple loci in the model. In this chapter, we will review all these aspects, illustrating them with a few examples. Finally, we will present the most recent developments in the domain.