As a part of a major AGARD project concerning thermophysical properties of solid materials at high temperatures, cooperative thermal expansion measurements up to 1000°C were carried out using pushrod techniques. Sixteen laboratories have contributed results. Parallel samples having the same history were used for all tests. Materials investigated were very pure gold and platinum, an austenitic alloy, alumina, and three types of graphite. From the mean values of the individual contributors, band widths of group scatter as well as group averages were derived. In addition, error analyses were carried out to show if the errors involved were predominantly of a systematic or of a non‐systematic nature. For gold and platinum, the absolute accuracy of the group averages amounting to 0.005 per cent expansion was found to be no worse than that of individual averages by X‐ray and optical measurements. Of the technical materials investigated, the austenitic alloy, alumina, and a special ultrafine‐grain graphite were found to be very suitable materials. In the case of alumina, a correction of the originally submitted individual averages by calibrating the apparatuses with platinum was found to decrease the band width of group scatter to any significant extent but again the group averages remained unchanged. Regarding the accuracy obtained by the pushrod dilatometers used in this cooperative program it was observed that it is possible for an individual calibration to be less reliable than a statistical evaluation of a group effort. This surprising result indicates that the individual measurement conditions differ from those of the individual calibration runs. For the pushrod dilatometer reference system, up to 900°C silica glass was found to be acceptable but not better than alumina. Ultra‐fine‐grain graphite looks promising for the same application.