To explore the difference of the proximal femoral shortening (PFS) between the third generation of Gamma nail (TGN) and the proximal femoral nail anti-rotation (PFNA) in treating intertrochanteric fracture of femur. The clinical data of 158 patients with intertrochanteric fracture of femur who were treated with TGN internal fixation or PFNA internal fixation between January 2014 and December 2015 were retrospectively analysed. The patients were divided into TGN group (69 cases) and PFNA group (89 cases) according to surgical operation. There was no significant difference in gender, age, bone mineral density, causes of injury, AO/Association for the Study of Internal Fixation (AO/ASIF) classification, accompanied disease, and the time from injury to operation between 2 groups ( P>0.05). The result of fracture reduction was divided into 3 types: positive medial cortex support, neutral position cortex support, and negative medial cortex support according to the method of Chang et al. At 18 months postoperatively, bilateral hip anteroposterior X-ray films were taken to measure horizontal PFS values (marked as X), vertical PFS values (marked as Y), and calculate the total PFS values (marked as Z). The PFS values were divided into 4 grades according to the criteria (≤1.0 mm, 1.0-4.9 mm, 5.0-9.9 mm, and ≥10.0 mm), and the constituent ratio was calculated and compared between 2 groups. The X, Y, and Z values and the collodiaphyseal angles of 2 groups at 18 months postoperatively were compared. The X, Y, and Z values of 2 groups of patients with failed fixation and normal healing within 18 months after operation were recorded and compared. The X, Y, and Z values of 2 groups of the patients with different cortex support types were also compared. There were 34 cases of positive medial cortex support, 30 cases of neutral position cortex support, and 5 cases of negative medial cortex support in TGN group, and there were 45, 33, and 11 cases in PFNA group respectively, showing no significant difference between 2 groups ( Z=-1.06, P=0.29). All patients were followed up 18 months after operation. At 18 months after operation, the constituent ratios of PFS values (X, Y, Z) had significant differences between 2 groups ( P<0.05). The patients of shortening of 1.0-4.9 mm and 5.0-9.9 mm were obviously more in TGN group than in PFNA group; the patients of shortening of ≥10.0 mm were obviously more in PFNA group than in TGN group. There were significant differences in X, Y, and Z values between 2 groups ( P<0.05), but no significant difference of the collodiaphysial angle was found between 2 groups ( t=0.47, P=0.64). Six cases of internal fixation failed in TGN group and PFNA group respectively within 3 months after operation, and there was no significant difference of X, Y, and Z values between failed fixation and normal healing patients within 2 groups ( P>0.05). When the reposition effect was the positive medial cortex support, the X, Y, and Z values were significantly lower in TGN group than in PFNA group ( P<0.05); but no significant difference was found between 2 groups when the reposition effect was the neutral position cortex support or negative medial cortex support ( P>0.05). At 18 months after operation, the X, Y, and Z values of the negative medial cortex support patients were significantly higher than those of the positive medial cortex support or the neutral position cortex support patients within 2 groups ( P<0.05). PFS is a common complication of the intertrochanteric fracture of the femur after internal fixation. During operation, the selection of internal fixation should be based on the results of intraoperative reduction. TGN should be applied to reduce PFS if positive medial cortex support happened.