Background/Objectives: After tooth extraction the crest may close before the socket has mineralized, leaving a covered socket residuum (CSR). Platelet-rich fibrin (PRF) accelerated soft tissue closure of molar and premolar sockets in a randomized controlled trial. We asked whether PRF also changes the way these sockets fill with bone. Methods: Radiological analysis of a prospective, single-blind randomized controlled trial in which sockets were filled with solid PRF or left to heal unassisted. Cone beam computed tomography was acquired after extraction and after about three months. We assessed three-dimensional reconstructions for a CSR and for collapse of the walls and analysed registered scan pairs volumetrically with an observer-independent method. Results: Twenty-nine sockets were included (control n = 17, PRF n = 12). In all 21 sockets with an assignable three-dimensional reconstruction a CSR was found, and the walls moved inward in both groups. In 17 registered scan pairs, the median mineralized share of the original socket volume was 25% in the control and 51% in the PRF group (p = 0.07), and the non-mineralized volume 285 and 144 mm³ (p = 0.046). After adjustment for tooth type and follow-up interval, the mineralized share was 18 percentage points higher with PRF (p = 0.03). An earlier Fiji-based analysis of the same sockets showed the same group difference (p = 0.03). Every socket retained a coherent non-mineralized space. Conclusions: PRF did not prevent socket collapse or a CSR in premolar and first molar sockets, although it shifted early healing toward mineralization. Where the socket volume is to be preserved, a space-maintaining material is needed, with PRF as its carrier.