Placental infarction or abruption, recurrent pregnancy loss and pre-eclampsia are thought to arise due to defects within the placental vascular bed. Deficiencies of vitamin B12 and folate, or other abnormalities within the methionine-homocyst(e)ine pathway have been implicated in the development of such placental diseases. We conducted a systematic literature review to quantify the risk of placental disease in the presence of these metabolic defects. Studies were identified through OVID Medline between 1966 and February 1999. Terms relating to the measurement of vitamin B12, folic acid, methylenetetrahydrofolate reductase or homocyst(e)ine were combined with those of pre-eclampsia, placental abruption/infarction or spontaneous and habitual abortion. Human studies comprising both cases and controls and published in the English language were accepted. Their references were explored for other publications. Data were abstracted on the matching of cases with controls, the mean levels of folate, B12 or homocyst(e)ine in each group or the frequency of the homozygous state for the thermolabile variant of methylenetetrahydrofolate reductase. The definition of 'abnormal' for each exposure was noted and the presence or absence of the exposure of interest for each outcome was calculated as an absolute rate with a 95 per cent confidence interval. The crude odds ratios were calculated for each study and then pooled using a random effects model. Eighteen studies were finally included. Eight studies examined the risk of placental abruption/infarction in the presence of vitamin B12 or folate deficiency, or hyperhomocyst(e)inaemia. Folate deficiency was a prominent risk factor for placental abruption/infarction among four studies, though not statistically significant (pooled odds ratio 25.9, 95 per cent CI 0.9-736.3). Hyperhomocyst(e)inaemia was also associated with placental abruption/infarction both without (pooled odds ratio 5.3, 95 per cent CI 1.8-15.9) and with methionine loading (pooled odds ratio 4.2, 95 per cent CI 1.2-15.0), as was the homozygous state for methylenetetrahydrofolate reductase (pooled odds ratio 2.3, 95 per cent CI 1.1-4.9). Vitamin B12 deficiency was not a demonstrable risk factor. Eight studies examined blood levels among women with spontaneous abortion or recurrent pregnancy loss. The pooled odds ratios were 3.4 (95 per cent CI 1.2-9.9) for folate deficiency, 3.7 (95 per cent CI 0.96-16.5) for hyperhomocyst(e)inaemia following methionine challenge, and 3.3 (95 per cent CI 1.2-9.2) for the methylenetetrahydrofolate reductase mutation. Five case-control studies examined the relationship between pre-eclampsia and abnormal levels of vitamin B12, folate, homocyst(e)ine or methylenetetrahydrofolate reductase. Folate deficiency was not an associated risk factor (odds ratio 1.2, 95 per cent CI 0.5-2.7), but hyper-homocyst(e)inaemia was (pooled odds ratio 20.9, 95 per cent CI 3.6-121.6). Similarly, homozygosity for the methylenetetrahydrofolate reductase thermolabile variant was associated with a moderate risk of preeclampsia (odds ratio 2.6, 95 per cent CI 1.4-5.1). Some pooled data were associated with significant statistical heterogeneity, however. There is a general agreement among several observational studies that folate deficiency, hyperhomocyst(e)inaemia and homozygosity for the methylenetetrahydrofolate reductase thermolabile variant are probable risk factors for placenta-mediated diseases, such as pre-eclampsia, spontaneous abortion and placental abruption. Vitamin B12 deficiency is less well defined as an important risk factor. Due to the limited quality of these data, including insufficient matching of cases with controls, and possible laboratory measurement bias relating to pregnancy, prospective studies are needed to confirm these findings and guide future preventative and therapeutic research. Copyright 1999 Harcourt Publishers Ltd.