The effect of hardware and software on the quality of Cerec all-ceramic partial crowns was investigated in this cross-sectional study. Partial crowns (n = 818) had been adhesively placed in 496 patients between 1993 and 1997 using Cerec 1 and Cerec 2 units (groups 1 and 2) as well as Cerec 2 with wall-spacing software (group 3). From each group, 25 randomly selected partial crowns were evaluated using modified United States Public Health Service (USPHS) criteria. Of these, 12 were randomly selected in each group, replicas taken and examined in a scanning electron microscope for marginal interfacial width and for continuous margin adaptation. Interfacial width of group 1 (308 +/- 95 micro m) was significantly larger than those of groups 2 (243 +/- 48 micro m) and 3 (207 +/- 63 micro m). Continuous margin adaptation at the tooth-luting composite and luting composite-restoration interfaces showed only minor differences in groups 1 (94.5 +/- 8% and 95.5 +/- 2%), 2 (98.1 +/- 1% and 97.5 +/- 1.4%) and 3 (96.8 +/- 3% and 96.8 +/- 2%). Pooled clinical rating was excellent or good at 97% for all groups, indicating acceptable restoration quality except for one breakage in group 1.