To explore the clinical value and safety of unilateral pedicle screw fixation combined with contralateral translaminar facet screw fixation and interbody fusion by muscle-splitting approach treatment of recurrent lumbar disc herniation. The clinical data of 51 patients with recurrent lumbar disc herniation treated from June 2012 to December 2017 were retrospectively analyzed. There were 32 males and 19 females, aged 34 to 64 years with an average of (51.11± 7.28) years. Lesions invoved L4,5 in 38 cases and L5S1 in 13 cases. All patients had a history of lower back pain and radiation pain of lower limbs(3 bilateral and 48 unilateral)and underwent unilateral pedicle screw combined with contralateral translaminar facet screw fixation and interbody fusion, among which 24 patients were treated through median incision approach (median incision group);other 27 patients were treated through muscle-splitting approach with channel-assisted exposure(muscle-splitting approach group). Operation time, intraoperative blood loss, postoperative drainage and incision length of the two groups were recorded. Visual analogue scale (VAS) was used to score the pain of lumbar incision at 72 h after operation, and JOA low back pain scoring system was used to evaluate the lumbar function preoperatively and at final follow-up. Imaging data were analyzed, including the changes in the height of intervertebral space of diseased segment before operation, 3 to 5 days after operation, and at final follow-up;Cobb angle changes in the coronal and sagittal planes of lumbar spine preoperatively and at final follow-up;multifidus area and multifidus fatty tissue deposition grade before and 12 months after operation; postoperative pedicle screw and laminar process screw position and intervertebral fusion condition. The complications of the two groups were compared. There was no statistical difference in operation time between two groups (P>0.05). Muscle-splitting approach group was better than median incision group in light of incision length, intraoperative blood loss and postoperative drainage volume (P<0.05). VAS score of lumbar incision pain at 72 h after operation was 1.61±0.54 in median incision group and 0.76±0.28 in muscle-splitting approach group(P<0.05). All patients were followed up for 12 to 84 (43.50±15.84) months. At final follow-up, the JOA scores of the two groups were significantly improved compared with those before operation(P<0.05). The rate of pedicle screw malposition was 6.25%(3/48) in medianincision group and 9.26%(5/54) in muscle-splitting approach group, there was no statistically significant difference between two groups (P>0.05). Rate of translaminar facet screw malposition in median incision group (12.50%) was significant less than the muscle-splitting approach group (18.52%)(P< 0.05). The height of the intervertebral space of the two groups was significantly restored 3 to 5 days after operation (P<0.05), and there was also a significant loss of height at final follow-up (P<0.05). At final follow-up, the balance of lumbar coronal plane and sagittal plane in two groups were improved very well (P<0.05). The comparison of the area and grade of the multifidus muscle in two groups 12 months after operation showed that obvious damage to the multifidus muscle were present in the median incision, while the multifidus muscle was less damaged by muscle-splitting approach (P<0.05). The fusion rate was 91.7%(22/24) in the median incision group and 92.6%(25/27) in muscle-splitting approach group(P>0.05). In median incision group, there were 1 case of intraoperative pedicle entry point fracture, 1 case of intraoperative dural tear and 1 case of postoperative nerve root injury;in muscle-splitting approach group, there were 1 case of intraoperative pedicle entry point fracture, 2 cases of intraoperative dural tear, 1 case of postoperative nerve root injury, 2 cases of incision epidermal necrosis and 1 case of poor incision healing. Nerve root injuries in the two groups were caused by incorrect positions of pedicle screws, the screws were immediately adjusted upon discovery. The nerve root symptoms were completely recovered 3 and 6 months after surgery. No incision infection was occurred in two groups. During the follow-up, no pedicle screw and laminar facet screw were loosened, displaced, broken, or intervertebral fusion cage moved forward and backward. The complication rate of 25.93% in muscle-splitting approach group was higher than 12.50% in the median incision group (P<0.05). Muscle-splitting approach is feasible for thetreatment of recurrent lumbar disc herniation with pedicle screw fixation combined with contralateral translaminar facet screw fixation and interbody fusion. Compared with the median incision approach, the muscle-splitting approach has the advantages of small incision, less trauma, less bleeding, rapid recovery. Also it can protect multifidus and do not increase the incidence of serious complications. Thus, it can be used as a choice for fixation and fusion of recurrent lumbar disc herniation.