OBJECTIVES: The use of lithium disilicate dental ceramics is increasing in dentistry and previous reports have suggested that they may have greater biological risks than previously thought. We tested a hypothesis that composition and processing influence the biological properties of these ceramics. METHODS: The cytotoxicity of two machined and three pressed lithium disilicate materials (n=6) were tested in vitro using mouse fibroblasts in direct contact with the materials for 72h. Cellular response was estimated by mitochondrial succinate dehydrogenase activity (MTT method). Mitochondrial activity was expressed as a percentage of Teflon controls, then compared to Teflon using 2-sided t-tests (alpha=0.05). Polished materials were aged in artificial saliva and tested for cytotoxicity periodically over 6 weeks, then were repolished (320grit SiC paper), aged and tested again for 4 weeks. RESULTS: All materials significantly (50-70%) suppressed cellular mitochondrial activity in the initial week, but suppression decreased by 25-30% over the next 2 weeks. In weeks 4 and 6 some materials exhibited a cytotoxic 'relapse' of 10-20%. The cytotoxic response was no different for machined or pressed materials, but the presence of ZnO had at least an association with longer-term cytotoxicity and relapse. Repolishing to 320grit did not increase cytotoxicity significantly. SIGNIFICANCE: Our results suggest that lithium disilicates are not biologically inert, and that many have a similar cytotoxicity dynamic regardless of small differences in composition or processing.