Purpose The aim of this study was to assess the effect of connection type and impression technique on the accuracy of fit of implant‐supported fixed complete‐arch dental prostheses (IFCDPs).

Materials and Methods An edentulous mandibular cast with five implants was fabricated to serve as master cast (control) for both implant‐ and abutment‐level baselines. A titanium one‐piece framework for an IFCDP was milled at abutment level and used for accuracy of fit measurements. Polyether impressions were made using a splinted and non‐splinted technique at the implant and abutment level leading to four test groups, n = 10 each. Hence, four groups of test casts were generated. The impression accuracy was evaluated indirectly by assessing the fit of the IFCDP framework on the generated casts of the test groups, clinically and radiographically. Additionally, the control and all test casts were digitized with a high‐resolution reference scanner (IScan D103i, Imetric, Courgenay, Switzerland) and standard tessellation language datasets were generated and superimposed. Potential correlations between the clinical accuracy of fit data and the data from the digital scanning were investigated. To compare the accuracy of casts of the test groups versus the control at the implant and abutment level, Fisher's exact test was used.

Results Of the 10 casts of test group I (implant‐level splint), all 10 presented with accurate clinical fit when the framework was seated on its respective cast, while only five of 10 casts of test group II (implant‐level non‐splint) showed adequate fit. All casts of group III (abutment‐level splint) presented with accurate fit, whereas nine of 10 of the casts of test group IV (abutment‐level non‐splint) were accurate. Significant 3D deviations (P < 0.05) were found between group II and the control. No statistically significant differences were found between groups I, III, and IV compared with the control. Implant connection type (implant level vs. abutment level) and impression technique did affect the 3D accuracy of implant impressions only with the non‐splint technique (P < 0.05).

Conclusion For one‐piece IFCDPs, the implant‐level splinted impression technique showed to be more accurate than the non‐splinted approach, whereas at the abutment‐level, no difference in the accuracy was found.