Statement of problem. The effect of combining new and recast metal on the compositional stability of the alloy used and accuracy of the final restoration is not known. Purpose. This study characterized the elemental compositional stability of as-received and recast type III gold alloy. The effect of combining these alloys on the marginal accuracy of complete cast crowns also was evaluated. Material and methods. Sixty standardized type III gold alloy complete crowns were made on epoxy resin dies duplicated from a metal master die with conventional laboratory techniques. Three casting protocols (n=20 each) were established in relation to the proportion of as-received and recast gold alloys: Group A (100% as-received metal), Group B (50% wt new metal, 50% wt once recast metal), and Group C ( 100% once recast metal). An elemental analysis was performed at 3 sites along the cast crown (cervical, middle, and occlusal) with x-ray energy-dispersive spectroscopy. Marginal accuracy was measured with a microscope before and after cementation. Results were subjected to 1-way analysis of variance followed by the Ryan-Einot-Gabriel-Welsch multiple-range test at a 1% level of significance. Results. Elemental composition was significantly different among the casting groups (P<.001). The mean weight percentage values were 72.4% to 75.7% Au, 4.5% to 7.0% Pd, 10.7% to 11.1% Ag, 7.8% to 8.4% Cu, and 1.0% to 1.4% Zn. Statistically but not clinically significant differences also were found fur marginal accuracy. The marginal discrepancy was less than 25 μm for all casting conditions, with the lowest value recorded for Group A (7 μm), the highest for Group C (12 μm), and an intermediate value for Group B (9 μm) specimens. Conclusion. Recasting type III gold alloys may adversely affect the marginal accuracy of complete cast crowns. In this study, however, such effects were not clinically significant.