Synopsis This is the third of three papers presenting the results of an investigation into the transient thermal strain behaviour of concrete during the first heat cycle to 600°C under load. The strains measured during the first cooling of concrete from 600°C are presented here, together with the results of residual compressive strength and elastic modulus tests. These are analysed with reference to the complete thermal cycle and constituent materials property data already given in the two previous papers. Strains during first cooling were primarily influenced by the thermal strain of the coarse aggregate and were unaffected by initial moisture content (as-cast, air-dry or pre-dried at 105°C), heating rate (0·2 or 1°C/min) or concrete age (1 year or 9 years). The applied load level did not affect the cooling strains of concretes with the more thermally stable aggregates of basalt or sintered pulverized-fuel ash, or the siliceous gravel concrete despite the extensive cracking it suffered during heating. This indicated the absence of transient creep during cooling and also the insignificance of any thermal stress effects for the specimen size and cooling rates used, The thermal expansion of limestone aggregate during heating contained an irreversible component which created sufficient differential cooling strain between cement paste and aggregate to cause cracking on cooling. The degree of cracking reduced with increasing levels of applied compressive load (0–30% of the initial unheated strength). The residual strain results represent the sum of all irrecoverable contractions and expansions, during the thermal cycle. These include shrinkage and load-induced thermal strain (LITS) during first heating and the effects of any cracking on cooling. From this work, the residual contractions of the loaded specimens can be successfully predicted by a simple superposition of all irrecoverable strains developed during the thermal cycle. The effect of temperature upon elasticity was dependent upon whether or not the concrete was loaded during the thermal treatment. For loaded concretes tested hot after a heating-cooling-heating cycle, the elastic strain response changed little between 100°C and 600°C compared with the free thermal strain (FTS) and LITS. For concretes tested cold after heating under load but cooling unloaded, the elastic strains increased with temperature level. Residual compressive strength of all the concretes was a function of the load level during the thermal cycle. The residual strength of specimens initially heated at 0.2°C/min was less than that of specimens heated at 1.0°C/min. This indicated that longer time at temperature was more influential than possible structural' effects due to spatial temperature variations in the specimens during heating.