Environmental risk-management decisions in the U.S. involving potential exposures to methylmercury currently use a reference dose (RfD) developed by the U.S. Environmental Protection Agency (USEPA). This RfD is based on retrospective studies of an acute poisoning incident in Iraq in which grain contaminated with a methylmercury fungicide was inadvertently used in the baking of bread. The exposures, which were relatively high but lasted only a few months, were associated with neurological effects in both adults (primarily paresthesia) and infants (late walking, late talking, etc.). It is generally believed that the developing fetus represents a particularly sensitive subpopulation for the neurological effects of methylmercury. The USEPA derived an RfD of 0.1 microg/kg/day based on benchmark dose (BMD) modeling of the combined neurological endpoints reported for children exposed in utero. This RfD included an uncertainty factor of 10 to consider human pharmacokinetic variability and database limitations (lack of data on multigeneration effects or possible long-term sequelae of perinatal exposure). Alcoa signed an Administrative Order of Consent for the conduct of a remedial investigation/feasibility study (RI/FS) at their Point Comfort Operations and the adjacent Lavaca Bay in Texas to address the effects of historical discharges of mercury-containing wastewater. In cooperation with the Texas Natural Resource Conservation Commission and USEPA Region VI, Alcoa conducted a baseline risk assessment to assess potential risk to human health and the environment. As a part of this assessment. Alcoa pursued the development of a site-specific RfD for methylmercury to specifically address the potential human health effects associated with the ingestion of contaminated finfish and shellfish from Lavaca Bay. Application of the published USEPA RfD to this site is problematic; while the study underlying the RfD represented acute exposure to relatively high concentrations of methylmercury, the exposures of concern for the Point Comfort site are from the chronic consumption of relatively low concentrations of methylmercury in fish. Since the publication of the USEPA RfD, several analyses of chronic exposure to methylmercury in fish-eating populations have been reported. The purpose of the analysis reported here was to evaluate the possibility of deriving an RfD for methylmercury, specifically for the case of fish ingestion, on the basis of these new studies. In order to better support the risk-management decisions associated with developing a remediation approach for the site in question, the analysis was designed to provide information on the distribution of acceptable ingestion rates across a population, which could reasonably be expected to be consistent with the results of the epidemiological studies of other fish-eating populations. Based on a review of the available literature on the effects of methylmercury, a study conducted with a population in the Seychelles Islands was selected as the critical study for this analysis. The exposures to methylmercury in this population result from chronic, multigenerational ingestion of contaminated fish. This prospective study was carefully conducted and analyzed, included a large cohort of mother-infant pairs, and was relatively free of confounding factors. The results of this study are essentially negative, and a no-observed-adverse-effect level (NOAEL) derived from the estimated exposures has recently been used by the Agency for Toxic Substances and Disease Registry (ATSDR) as the basis for a chronic oral minimal risk level (MRL) for methylmercury. In spite of the fact that no statistically significant effects were observed in this study, the data as reported are suitable for dose-response analysis using the BMD method. Evaluation of the BMD method used in this analysis, as well as in the current USEPA RfD, has demonstrated that the resulting 95% lower bound on the 10% benchmark dose (BMDL) represents a conservative estimate of the traditional NOAEL, and that it is superior to the use of "average" or "grouped" exposure estimates when dose-response information is available, as is the case for the Seychelles study. A more recent study in the Faroe Islands, which did report statistically significant associations between methylmercury exposure and neurological effects, could not be used for dose-response modeling due to inadequate reporting of the data and confounding from co-exposure to polychlorinated biphenyls (PCBs). BMD modeling over the wide range of neurological endpoints reported in the Seychelles study yielded a lowest BMDL for methylmercury in maternal hair of 21 ppm. This BMDL was then converted to an expected distribution of daily ingestion rates across a population using Monte Carlo analysis with a physiologically based pharmacokinetic (PBPK) model to evaluate the impact of interindividual variability. The resulting distribution of ingestion rates at the BMDL had a geometric mean of 1.60 microg/kg/day with a geometric standard deviation of 1.33; the 1st, 5th, and 10th percentiles of the distribution were 0.86, 1.04, and 1.15 microg/kg/day. In place of the use of an uncertainty factor of 3 for pharmacokinetic variability, as is done in the current RfD, one of these lower percentiles of the daily ingestion rate distribution provides a scientifically based, conservative basis for taking into consideration the impact of pharmacokinetic variability across the population. On the other hand, it was felt that an uncertainty factor of 3 for database limitations should be used in the current analysis. Although there can be high confidence in the benchmark-estimated NOAEL of 21 ppm in the Seychelles study, some results in the New Zealand and Faroe Islands studies could be construed to suggest the possibility of effects at maternal hair concentrations below 10 ppm. In addition, while concerns regarding the possibility of chronic sequelae are not supported by the available data, neither can they be absolutely ruled out. The use of an uncertainty factor of 3 is equivalent to using a NOAEL of 7 ppm in maternal hair, which provides additional protection against the possibility that effects could occur at lower concentrations in some populations. Based on the analysis described above, the distribution of acceptable daily ingestion rates (RfDs) recommended to serve as the basis for site-specific risk-management decisions at Alcoa's Point Comfort Operations ranges from approximately 0.3 to 1.1 microg/kg/day, with a population median (50th percentile) of 0.5 microg/kg/day. By analogy with USEPA guidelines for the use of percentiles in applications of distributions in exposure assessments, the 10th percentile provides a reasonably conservative measure. On this basis, a site-specific RfD of 0.4 microg/kg/day is recommended.