High-temperature stress during the reproductive growth stages of wheat has a devastating effect on yield. Temperatures that exceed 28°C during pollen formation and anthesis can decrease yield by 4% for every 1°C rise in average seasonal temperature. A three-tiered method of phenotyping was used to screen large numbers of lines for heat tolerance. This three-tiered approach initially evaluated large numbers of lines in dates of sowing experiments, with the responses confirmed on an optimally sown subset using portable field-based heat chambers that impose a heat stress 5°C above ambient temperature. Those lines showing a consistent response in the chambers were then evaluated under controlled conditions in the greenhouse. A training population of 2,500 wheat genotypes, specifically selected on the basis of putative heat tolerance, was evaluated between 2016 – 18 at the I.A. Watson Grains Research Centre, Narrabri, NSW, Australia, using this methodology. A subset of 200 lines with high genomic estimated breeding values that best captured the genetic diversity within the training population was maintained across years and phenotyped at multiple sites (Western Australia and Victoria, Australia). Genomic predictions, trained at Narrabri, were validated for accuracy at other sites and environmental covariates incorporated in the models to increase the prediction accuracies for yield, screenings and other traits. This phenotyping approach underpins our genomic selection program for heat tolerance in wheat.