Purpose To explore the pathological changes in optic nerve injury models under varying forces. Methods The rats were classified into 4 groups: sham operation (SH), 0.1, 0.3, and 0.5 N. Modeling was performed using the lateral optic nerve pulling method. Seven days after modeling, Brn3a immunofluorescence was used to detect retinal ganglion cell (RGC) number, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining was used to detect RGC apoptosis, and flash visual evoked potential (FVEP) was used to detect the optic nerve function on days 1, 3, and 7 after modeling. In addition, LC3 II and P62 expression levels in retinal tissues were detected by western blotting to observe the changes in autophagy levels. Results RGC number decreased 7 d after modeling, and it showed a downward trend with increasing damaging force. The number of apoptotic RGCs in ganglion cell layer in the 0.3 and 0.5 N groups was increased and was higher than that in the 0.1 N group. The difference in FVEP of rats in each group was mainly reflected in the P2 peak latency. LC3 II and P62 expression levels in retinal tissue of 0.3 and 0.5 N groups were higher than those of the SH and 0.1 groups; however, the difference between the 0.1 N and SH groups was not statistically significant. Conclusion Precisely controlling the force of the optic nerve clamping injury model is necessary because different forces acting on the optic nerve will lead to differences in the loss of optic neurons, the conduction function of the optic nerve, and autophagy level in retinal tissues.