In cost-effectiveness analysis and contemporary treatment planning strategies, the postponement of placement of cast crowns plays an important role. Extensive amalgam restorations that involve the rebuilding of cusps and the provision of auxiliary retention are thought to make this postponement possible. This study reports the long-term survival (100 months) of extensive amalgam restorations in a randomized, controlled clinical trial. The operational hypothesis was that the type of retention and the operator have a long-term influence on the survival and clinical functioning of extensive amalgam restorations. Three hundred extensive amalgam restorations were placed by three operators in molar teeth in which one or more cusps were reconstructed. Five different auxiliary retention methods were used for retention of these restorations. Since the purpose of extensive amalgam restorations is considered to be two-fold (to restore a broken-down molar to function acceptably as an independent restoration and to create a substructure for subsequent crown construction), survival was assessed at different levels. The survival rate of extensive amalgam restorations as an independent restoration was 88+/-2%. The functional survival rate (as an independent restoration or as a substructure) was 92+/-2%. The influence of experimental variables ('retention method' and 'operator') and background variables ('tooth type', 'extension of extensive amalgam restorations', and 'age of patient') on the survival was analyzed by Log Rank and Breslow tests. The analyses revealed that there were no statistically significant influences on the survival rates except for the variable 'age of patient' (p < or = 0.05). Extensive amalgam restorations were more prone to failure in the group of older patients than in the group of younger ones. It is concluded that the clinical survival of extensive amalgam restorations is independent from several clinical variables in the study.