Bone characteristics, the implant macrogeometry, and the drilling technique are considered the main important factors to obtain a good implant primary stability (PS). Indeed, although it is known that implant placement in poor bone sites increases the possibility of implant failure, several surgical procedures have been proposed to improve PS, such as site under-preparation. Hence, this in vitro study aimed to evaluate the insertion torque (IT), removal torque (RT), and resonance frequency analysis (RFA) of conical implants (3.3 and 4 × 13 mm) placed in under-prepared sites on 10 and 20 pounds per cubic foot (PCF) density polyurethane sheets (simulating a D3 and D2 bone, respectively) with and without a cortical sheet of 30 PCF in density (corresponding to a D1 bone). After using ANOVA or Kolmogorov–Smirnov test to elaborate data, the resulting IT and RT values were directly proportional to the polyurethane block densities (e.g., the lowest and highest IT values were 8.36 ± 0.52 Ncm in the 10 PCF density sheet and 46.21 ± 0.79 Ncm in the 20 PCF density sheet + cortical for 4 × 13 mm implants) and increased with the increasing amount of site under-preparation (the highest results for both implants were found with a 2.2 mm under-preparation, showing a significantly higher IT with a p < 0.05 compared with others, especially in the highest-density sheets). Both implants inserted in the 20 PCF density block + cortical with all under-preparation protocols exhibited significantly higher RFA values (p < 0.05–0.0001) compared with the corresponding ones in the 10 PCF block. Moreover, 3.3 × 13 mm implants showed the same results comparing the 20 PCF block and the 10 PCF block + cortical. In conclusion, in this in vitro study using low-density polyurethane blocks, the under-preparation of the implant insertion sites was shown to be effective in increasing implants’ PS.