The aim of the present study was to assess the clinical behaviour of cast gold restorations with respect to define a gold control against current and future ceramic and composite restorations. The study sample included 50 patients with 303 cast gold restorations. All restorations were cemented with a non-adhesive technique. A total of 303 restorations were evaluated clinically and radiographically using modified United States Public Health Service criteria. Restorations recorded as having an A- or a B-rating were defined as successful. Of the 303 restorations, 42 were judged as failures, which resulted in a failure rate of 13.8% for a mean observation time (+/- s.d.) of 18.7 ( +/- 9.5) years. The estimated Kaplan-Meier survival rates (s.e.) were 96.1% (+/- 1.1%) at 10 years, 87.0% (+/- 2.2%) at 20 years and 73.5% (+/- 5.4%) at 30 years. In total, biological reasons were counted 25 times in comparison to 17 technical reasons for those 42 failed cast gold restorations, with 17 secondary caries (40%) as the most common biological reason and with 13 retention losses (31%) as the most common technical reason. The endodontically treated tooth was exclusively identified as a risk factor. The restoration type (inlay versus onlay) did not influence the survival rate.