This paper documents a technique for abbreviating the climate data requirements for buildings thermal analysis programs and investigates the effect of various levels of abbreviation on program accuracy. The technique is incorporated in a computer program that reads a standard full year weather file and selects from each month’s hourly observations a segment representative of the month. The advantages and requirements of abbreviated data are first discussed followed by a brief description of the development of the technique. A series of initial tests using 96 hour segments is documented to show the technique’s effectiveness for envelope-dominated buildings, and a comparison of test runs using varied segment lengths reveals the advantages of using longer segments with more massive buildings.