OBJECTIVES: To determine the association between maternal occupational exposure to anaesthetic gases and risk of spontaneous abortion. METHODS: A meta-analysis was performed of published epidemiological studies identified from literature reviews, unsystematic perusal of reference lists of relevant publications, and two Medline searches (1984-92, keywords: anaesthetic gases; anaesthetics; anaesthetics, local; operating rooms; operating room nursing; pregnancy; abortion; 1985-92, keywords: anaesthetics; adverse effects; occupational exposure; anaesthesia, inhalation; operating room nursing; pregnancy; abortion). All peer reviewed studies were retained. Student theses were excluded, as were conference abstracts, unpublished material, and two studies in which data on paternal and maternal occupational exposures were pooled. The relative risk of spontaneous abortion was estimated. RESULTS: One study found no increase in risk of abortion when gases were scavenged or when the exposure to unscavenged gases was low. None of the studies included ambient gas sampling. 24 comparisons between exposed and unexposed women, obtained from 19 reports, were included. The overall relative risk was 1.48 (95% confidence interval (95% CI), 1.4 to 1.58). To test whether this result was influenced by the quality of the studies, the validity of the reviewed papers was rated on the basis of three criteria: appropriateness of the unexposed comparison group, control for non-occupational confounding variables, and response rate. The estimate of risk increased to 1.9 (95% CI, 1.72 to 2.09) when analysis was restricted to the six comparisons which were rated the most rigorous. CONCLUSIONS: Epidemiological studies based on data obtained in the prescavenging era indicate an increased risk of spontaneous abortion. The estimated increased risk was not diminished but rather increased by exclusion of the more methodologically flawed studies.