The current evidence on paediatric melanoma is heterogeneous, especially regarding the prognosis of different histological subtypes. We sought to systematically review the evidence on paediatric melanoma, highlighting the major sources of heterogeneity and focusing on available data on single patients. A systematic search was performed from 1948 to 25 January 2021. Only studies reporting at least one case of cutaneous melanoma in patients aged ≤18 years were included. Unknown primary and uncertain malignant melanomas were excluded. Three couples of authors independently performed title/abstract screening and two different authors reviewed all the relevant full texts. The selected articles were manually cross‐checked for overlapping data for qualitative synthesis. Subsequently data on single patients were extracted to perform a patient‐level meta‐analysis. PROSPERO registration number: CRD42021233248. The main outcomes were melanoma‐specific survival (MSS) and progression‐free survival (PFS) outcomes. Separate analyses were done of cases with complete information on histologic subtype, focusing on superficial spreading (SSM), nodular (NM) and spitzoid melanomas, as well as of those classified as de‐novo (DNM) and acquired or congenital nevus‐associated melanomas (NAM). The qualitative synthesis covered 266 studies; however, data on single patients were available from 213 studies including 1002 patients. Among histologic subtypes, NM had a lower MSS than both SSM and spitzoid melanoma, and a lower PFS than SSM. Spitzoid melanoma had a significantly higher progression risk than SSM and trended toward lower mortality. Focusing on nevus‐associated status, DNM demonstrated better MSS after progression than congenital NAM, and no differences were highlighted in PFS. Our findings describe the existence of different biological patterns in paediatric melanoma. Specifically, spitzoid melanomas demonstrated intermediate behaviour between SSM and NM and showed a high risk of nodal progression but low mortality. This raises the question of whether spitzoid lesions are being over‐diagnosed as melanoma in childhood.