In order to determine the best laboratory procedure for titanium crown casting, a set of thermal expansion measurements and casting experiments were carried out using a casting machine (argon arc, pressure difference type) and three different investments, two conventional SiO2 based investments and a new Al2O3/MgO based investment. The thermal expansion measurements involved a cycle of heating and cooling. The relatively low mould temperatures recommended (200 degrees C) or chosen (350 degrees C) for the conventional investments provided zero or negative mould expansion for the compensation of metal shrinkage. Crowns made from these investments exhibited heavy reaction with the mould, and the common cleaning method of sand blasting appeared to be essential. This cleaning process, however, was not adequate for the assessment of casting accuracy as the short sand blasting time (15 s) rapidly altered the fit of the crowns. The metal reacted little with the new investment and the best compensation (0.15 mm discrepancy) for the metal shrinkage, as assessed 'as cast', was achieved when the investment was heated to 950 degrees C and then cooled to the recommended mould temperature (600 degrees C).