Purpose: To compare the fit and fracture load of temporary fixed partial prostheses fabricated by means of a conventional direct technique, milling, or 3D printing. Materials and Methods: A maxillary right first premolar and molar were prepared on a Frasaco cast, which was then duplicated 40 times. In total, 10 provisional three-unit fixed prostheses (Protemp 4, 3M) were made using the conventional technique with a putty mold. The 30 remaining casts were scanned to design a provisional restoration using CAD software. A total of 10 designs were milled (CEREC MC X5/shaded PMMA Disk, Dentsply Sirona), while the other 20 were 3D printed with one of the two 3D printers (Asiga UV MAX or Nextdent 5100, C&B, Nextdent). Internal and marginal fit were examined using the replica technique. Next, the restorations were cemented on their respective casts and loaded until fracture using a universal testing machine. The location and propagation of the fracture were also evaluated. Results: 3D printing demonstrated the best internal fit. Nextdent (median internal fit: 132 μm) was significantly better compared to the milled (median internal fit: 185 μm; P = .006) and conventional restorations (median internal fit: 215 μm; P < .001), while the fit of Asiga (median internal fit: 152 μm) was only significantly better than the conventional restorations (P < .012). The lowest marginal discrepancy was found for the milled restorations (median marginal fit: 96 μm), but this was only significant when compared to the conventional group (median internal fit: 163 μm; P < .001). The conventional restorations demonstrated the lowest fracture load (median fracture load: 536 N), which was only significant when compared to Asiga (median fracture load: 892 N; P = .003). Conclusions: Within the present in vitro study’s limitations, CAD/CAM demonstrated superior fit and strength compared to the conventional technique.