The research carried out at the Canberra phytotron, CERES, during the first twenty years of its operation is reviewed as a case study of the opportunities, problems and value of phytotron research. The climatic responses of a large number of species, both wild and domesticated, are examined as well as those of fungal and viral pathogens and rhizobial symbionts. Some aspects of the design of CERES are briefly considered, and a few aspects of the variation between plants in climatic response, with the C4 photosynthetic syndrome taken as a case history of the linkages between biochemistry, anatomy and response to climate. Research in ecology, forestry, horticulture, plant pathology, rhizobial microbiology, anatomy and ultrastructure—and specific aspects of biochemistry, plant nutrition and genetics—are reviewed to indicate the range of botanical disciplines which can profit from access to a phytotron. The ecological research at CERES, much of it with trees and perennial grasses, particularly highlights the value of combining field and phytotron studies. Work on seed proteins, hybrid vigor and plant growth regulators is also reviewed. The next part of the paper considers the effects of the major environmental factors, temperature, daylength, irradiance, atmospheric CO2 level and water stress, together with their interactions and the problems of correlating phytotron and field responses. The effect of these factors on the various processes of growth and development are considered, and the stages of development most sensitive to them. An evolutionary perspective on yield potential is then discussed, leading to a consideration of photosynthesis, translocation, partitioning and storage organ growth as yield-determining processes. The final part considers the uses of phytotrons in agricultural, horticultural and forestry research, in terms of examples from work at CERES on the manipulation of breeding systems, the clarification of plant breeding objectives, the use of phytotron conditions for selection, the prediction of adaptation and spread, and models for yield prediction and pest management. Several areas requiring more attention are identified, and some conclusions are drawn on the value of access to a phytotron such as CERES for botanical research of many kinds.