To evaluate and compare the clinical and histopathological profile of primary and recurrent orbital-periorbital plexiform neurofibromas (OPPN) in patients with neurofibromatosis type 1. We retrospectively evaluated 43 primary or recurrent neurofibroma (NF) specimens from 26 patients (2002 to 2018) at the King Khaled Eye Specialist Hospital, Saudi Arabia. Demographics, clinical presentation, and surgical intervention data were collected. Histopathological specimens were studied with hematoxylin-eosin, Alcian blue, and immunohistochemical markers; S-100, CD44, CD117, smooth muscle actin (SMA), neurofilament, and Ki-67. Of the 43 NFs specimens, 20 were primary and 23 recurrent tumors. For primary NF, the ratio of plexiform to the diffuse type was 13:7, however in recurrent tumors was 3:8 after the first recurrence, and 1:5 after multiple recurrences. Of the 17 patients with primary tumors that had paired recurrent tumors, 12/17 (70.6%) primary NFs were plexiform and 5/17 (29.4%) were diffuse. However, when tumors recurred, 13/17 tumors (76.5%) were diffuse and only 4/17 tumors (23.5%) had a plexiform pattern. The odds of a tumor having a diffuse pattern in recurrent NF was significantly higher than the plexiform pattern [OR = 7.8 (95% confidence interval 1.69:36.1) P = 0.008]. Primary plexiform NFs underwent an excision at a significantly younger age than the diffuse type. Recurrent NFs had significantly higher CD44, CD117, and neurofilament labeling (P = 0.02, P = 0.01 and P<0.001 respectively) but had significantly decreased Alcian blue, and S-100 labeling (P = 0.03, and P = 0.02 respectively) compared to primary tumors. SMA and Ki-67 proliferation index were not different between primary and recurrent NFs (P = 0.86, and P = 0.3 respectively). There appears to be a high risk for primary plexiform NFs to develop a diffuse histologic pattern when they recur. Immunohistochemical staining suggests a role of mast cells (CD117) and expression of infiltration makers (CD44) in the transformation of plexiform tumors to the diffuse phenotype.