Nerve destruction is central to the pathogenesis and clinical manifestation of leprosy reactions (LRs). However, pathological changes in dermal nerves in LRs have not been clearly elucidated. Hematoxylin and eosin (H&E) staining may fail to accurately identify fragmented nerves, and special stains may be required. We recruited 56 patients with clinically diagnosed LRs as cases and 30 patients with nonreactional leprosy as controls. Number and level of nerves, nerve edema, relation of granulomas to nerves, perineuritis, pattern of nerve involvement, and quantification of each nerve pattern were noted on H&E stain and S-100 immunostain on skin biopsy sections. Most of the cases were borderline tuberculoid (BT; 32.8%) and lepromatous leprosy (32.8%) types, whereas most controls were classified as BT (44.8%). We found greater dermal nerve infiltration, fragmentation, and destruction during reactions when compared with nonreactional leprosy (fragmented nerves on S-100: P <0.0001). Nerve fragmentation (P = 0.037), subcutaneous nerve involvement (P = 0.014), and severe nerve edema (P = 0.0005) were higher in type 2 reaction (T2R) compared with type 1 reaction (T1R; on S-100), mostly attributed to the higher number of severe T2Rs (n = 23/25) among our cases. Intact nerves were higher in downgrading T1R compared with upgrading T1R (P = 0.038 on H&E and P = 0.004 on S-100). Thus, the identification and quantification of different patterns of nerves using special stains, such as S-100, may shed more light on nerve fiber involvement and destruction in LRs and may help us predict the prognosis of such cases.