Provisional restorations were fabricated on a cast metal model and the resulting marginal discrepancies were measured. In the control group, the restorations were fabricated directly on the die and no coolant was applied. Five other methods of fabricating provisional restorations were evaluated. The mean marginal discrepancies were as follows: Control = 0.228 mm, In Situ (control plus coolant) = 0.196 mm, Removal (removed from die to polymerize) = 0.403 mm, On/Off (repeated removal/replacement) = 0.601 mm, Reline = 0.074 mm, and Indirect = 0.161 mm. The mean marginal discrepancy for the On/Off technique was significantly greater than that for all other techniques. The mean marginal discrepancy with the Removal technique was significantly greater than that with the Control, In Situ, Reline, and Indirect techniques. No significant differences were found between the Control, In Situ, Reline, and Indirect techniques.