Purpose: To evaluate the influence of abutment height and thermocycling on the retrievability of cemented implant crowns. Materials and Methods: Ninety tapered titanium abutments (6 degree taper, 4.3 mm diameter, 8.5 mm height) were shortened to 2, 3, or 4 mm, respectively. Ninety crowns were designed and manufactured using CAD/CAM techniques and laser sintering a CoCr alloy. The crowns were cemented either with a glass-ionomer, a polycarboxylate, or a composite resin cement followed by 3-day storage in demineralized water without thermocycling or 150-day storage with 37,500 thermal cycles. The force (in N) and the number of attempts needed to remove the crowns using a universal testing machine (UTM) or a clinically used removal device (Coronaflex) were recorded. Statistical analysis at a level of significance of P = .05 was conducted using the Kruskal-Wallis and Mann-Whitney U tests (Coronaflex) and three-way and two-way ANOVA, Tukey's HSD post hoc tests, and t tests (UTM). Results: Regardless of whether the crowns were retrieved with Coronaflex or UTM, the crowns cemented with the glass-ionomer cement were significantly easier to retrieve followed by the polycarboxylate and the resin cement, both of which differed significantly from each other (P = .001). With both retrieval methods, the cement, abutment height, and thermocycling were significantly influential (P = .0001). Significant interactions could be found for retrieval with UTM between the abutment height and thermocycling, between the cement and thermocycling, and between all three factors (P = .05). Conclusions: Glass-ionomer cement may be used for retrievable cementation of implant restorations, whereas polycarboxylate cement and especially composite resin cement should be used for a nonretrievable permanent cementation. Parole chiave: abutment, cement, crown, implant, retention, retrievability