Data sourcesThe Medline and Embase databases and hand searches in the journals International Journal of Paediatric Dentistry and European Archives of Paediatric Dentistry.Study selectionEnglish language cohort and case-control studies.Data extraction and synthesisStudy selection was carried out independently by two reviewers with data abstraction being conducted by a single reviewer and checked by a second reviewer. Risk of bias was assessed using a modified version of the Newcastle-Ottawa Scale (NOS). Adjusted (aOR) and unadjusted odds ratios (uOR), P-values and 95% confidence intervals (CI) were obtained from the studies. Meta-analysis was not conducted.ResultsTwenty-eight studies were included; 25 reported on MIH, three on hypomineralised second primary molars (HSPM). Nineteen of the studies were of cohort design (six prospective,13 retrospective) and nine were case controls. There was little evidence of an association between the most frequently investigated prenatal factors (smoking, maternal illness, maternal medication, maternal stress) and MIH. Similarly there was little evidence of an association between MIH and perinatal factors such as prematurity, low birth weight, caesarean delivery and birth complications. Early childhood illness, up to three or four years of age, was widely investigated, with six studies reporting a crude association. Associations between antibiotics, anti-asthma medication and breastfeeding were also evaluated. Only three studies looked at HSPM; one study suggested that maternal antibiotic use during pregnancy is unlikely to be associated with HSPM but maternal alcohol intake may be. Another study reported possible associations with a large number of factors, with perinatal factors and neonatal illness being most common, followed by prenatal factors.ConclusionsPrenatal and perinatal factors are infrequently associated with MIH. However, despite a lack of prospective studies, early childhood illness (in particular fever) appears to be associated with MIH. Further prospective studies that adjust for confounding based on biological principles, as well as genetic and epigenetic studies, are needed because the aetiology is likely to be multifactorial.