The aim of this study was to measure in vitro retention of cast gold crowns cemented with traditional and resin cements. Forty-eight human molars were prepared on a lathe to produce complete crown preparations with a consistent taper and split into six groups, eight crowns in each group. Crowns were cast in a high-gold alloy and then cemented. After 24 hours, the retention force (N) was recorded and mean values were analyzed by one-way analysis of variance and the Fisher post-hoc least significant difference (PLSD) multiple comparisons test (α = .05). Failure sites were examined under x100 magnification and recorded. Mean values (SD) for each group in increasing order of retention force were: Harvard Cement: 43 N (27), TempoCem: 59 N (16), PermaCem Dual: 130 N (42), RelyX Luting Cement: 279 N (26), Contax and PermaCem Dual: 286 N (38), and TempoCem with Contax and PermaCem Dual: 340 N (14). The Fisher PLSD interval (P = .05) for comparing cements was 29 N. Zincphosphate cement and provisional resin cements had the lowest retention forces. Resin cement with a bonding agent and the hybrid-ionomer cement had similar retention forces. Resin cement with a bonding agent applied after use of a provisional resin cement had a significantly higher retention force than the other cements tested.