Aims Presently, pathologists in the UK use diverse terminologies for the classification of sudden unexpected infant deaths, including ‘unascertained,’ ‘sudden unexpected death in infancy’ (SUDI) and ‘sudden infant death syndrome’ (SIDS). This study uses the Delphi method to investigate the views of paediatric pathologists on their use of these terms in order to determine areas of consensus.

Methods There were three Delphi rounds overall; in the final one, participants were asked to score each statement using a modified Likert scale (0–9). The scores were analysed using non-parametric statistics, and statements in which the median score was ≤3 or ≥7 (approximately 70% agreement) were considered to have reached ‘consensus agreement.’

Results Twenty-five of the 36 UK paediatric pathologists who were approached in the initial round contributed to all three rounds. There was consensus that ‘SIDS’ be used for unexplained sudden unexpected infant deaths that occurred during sleep. ‘Infancy’ was defined as up to 1 year of age, but there was no consensus regarding the lower age limit of SIDS. There was agreement that ‘SUDI’ be used for unexplained sudden infant deaths with a history of preceding illness, deaths with minor histological abnormalities of uncertain significance and co-sleeping-associated deaths. Most paediatric pathologists used ‘unascertained’ for findings suspicious of a non-natural cause of death. There was consensus that co-sleeping-associated deaths should be classified as ‘unascertained’ if parents had consumed alcohol or used drugs in the preceding 24 h.

Conclusions The areas of consensus relating to terminology around SUDI and SIDS should guide future use by pathologists. However, there remains a significant lack of agreement, suggesting that acceptable alternative terms be identified for infant deaths which remain unexplained following autopsy in whom there are no suspicious features; the authors propose that ‘unexplained SUDI,’ followed by a comment, may represent the most factually correct compromise.