C5 palsy is a frequent sequela of cervical decompression surgeries for cervical myeloradiculopathy. Although many researchers have suggested various risk factors, such as cord shifting and the correction of lordotic angles, the tethering of the C5 root beneath the narrow foramen is an independent risk factor for C5 palsy. In this study, we tried to investigate different techniques for foramen decompression with posterior cervical fusion and assess the incidence of C5 palsy with each technique depending on the order of foraminal decompression. A combined 540° approach with LMS and uncovertebrectomy was used in group 1. Group 2 combined a 540° approach with pedicle screws and posterior foraminotomy, while posterior approach only with pedicle screws and foraminotomy was used in group 3. For groups 2 and 3, prophylactic posterior foraminotomy was performed before laminectomy. Motor manual testing to assess C5 palsy, the Neck Disability Index (NDI) and the Japanese Orthopedic Association (JOA) scores were determined before and after surgery. Simple radiographs, MRI and CT scans, were obtained to assess radiologic parameters preoperatively and postoperatively. A total of 362 patients were enrolled in this study: 208 in group 1, 72 in group 2, and 82 in group 3. The mean age was 63.2, 65.5, and 66.6 years in groups 1, 2, and 3, respectively. The median for fused levels was 4 for the three groups. There was no significant difference between groups regarding the number of fused levels. Weight, height, comorbidities, and diagnosis were not significantly different between groups. Preoperative JOA scores were similar between groups (p = 0.256), whereas the preoperative NDI score was significantly higher in group 3 than in group 2 (p = 0.040). Mean JOA score at 12-month follow-up was 15.5 ± 1.89, 16.1 ± 1.48, and 16.1 ± 1.48 for groups 1, 2, and 3, respectively; it was higher in group 3 compared with group 1 (p = 0.008) and in group 2 compared with group 1 (p = 0.024). NDI score at 12 months was 13, 12, and 13 in groups 1, 2, and 3, respectively; it was significantly better in group 3 than in group 1 (p = 0.040), but there were no other significant differences between groups. The incidence of C5 palsy was significantly lower in posterior foraminotomy groups with pedicle screws (groups 2 and 3) than in LMS with uncovertebrectomy (group 1) (p < 0.001). Thus, preventive expansive foraminotomy before decompressive laminectomy is able to significantly decrease the root tethering by stenotic lesion, and subsequently, decrease the incidence of C5 palsy associated with posterior only or combined posterior and anterior cervical fusion surgeries. Additionally, such expansive foraminotomy might be appropriate with pedicle screw insertion based on biomechanical considerations.