Breeding for drought tolerance in grain crops is not a generic issue. Periods of drought vary in length, timing and intensity and different traits are important with different types of drought. The search for generic drought tolerance using single-gene transformations has been disappointing. It has typically concentrated on survival of plants suffering from severe water stress, which is rarely an important trait in crops. More promising approaches that target complex traits tailored to specific requirements at the different main stages of the life of a crop, during: establishment, vegetative development, floral development and grain growth are outlined. The challenge is to devise inexpensive and effective ways of identifying promising phenotypes with the aim of aligning them with genomic information to identify molecular markers useful to breeders. Controlled environments offer the stability to search for attractive phenotypes or genotypes in a specific type of drought. The recent availability of robots for measuring large number of plants means that large numbers of genotypes can be readily phenotyped. However, controlled environments differ greatly from those in the field. Devising pot experiments that cater for important yield-determining processes in the field is difficult, especially when water is limiting. Thus, breeders are unlikely to take much notice of research in controlled environments unless the worth of specific traits has been demonstrated in the field. An essential link in translating laboratory research to the field is the development of novel genotypes that incorporate gene(s) expressing a promising trait into breeding lines that are adapted to target field environments. Only if the novel genotypes perform well in the field are they likely to gain the interest of breeders. High throughput phenotyping will play a pivotal role in this process.