Several years ago, the State of California convened a workgroup of regulators and stakeholders to develop scientifically defensible minimum reporting limits (MRLs) for metals in drinking water. The workgroup was interested in lowering the reporting levels for the regulated metals to maximize occurrence information without jeopardizing the ability to use any methodology approved by the US Environmental Protection Agency (USEPA) for drinking water analysis. In order to validate its decisions, the workgroup carried out a large-scale interlaboratory study of performance on 16 metals using approved drinking water methodologies. The study used varying concentrations of each metal in two representative matrices (DI Water and a typical California drinking water matrix similar to State Project Water). Samples included blanks and blind spikes at different levels. In newer methods such as 314 (perchlorate), 317, and 326 (bromate), 515.4 (herbicides), and 526, 528, and 532 (UCMR List 2 methods), EPA has promulgated the concept of specifying precision/accuracy requirements at the MRL and maximum concentrations for blanks as basic method acceptance criteria. Under existing regulations, this approach provides a balance between laboratory capacity and laboratory performance. However, if the health risk of a given metal is significant at very low levels, a closer review is indicated to determine whether a lower reporting limit could be achieved by further restricting the number of approved methods, or by mandating specific extra QC performance criteria. This paper presents case studies of arsenic and barium, in which a close review of the data leads to different conclusions regarding achievable MRLs. The case studies evaluated the following parameters to determine MRLs that could be achieved routinely: the ability of each lab to obtain a result for the unspiked blank matrix that was below the target MRL (levels of 33%, 40%, and 50% of the target MRL were evaluated); the ability of each lab to obtain results for blind reagent water spikes (LFB) that were within specified target accuracy and precision at the desired MRL (two levels of LFB were provided per metal); the ability of each lab to obtain results for the blind matrix spikes that were within specified target accuracy and precision at the desired MRL (5 different spike levels bracketing the target MRL range were used); the ability of each lab to have a reported MDL that was consistent with its ability to accurately quantitate at a given target MRL (e.g. if a lab reported an MDL of 1 ppb, but could clearly not detect or quantify precisely at 1 ppb, it was considered a problem); and, all of these factors were evaluated for each approved method to determine whether MRLs might be limited by allowing a range of methods (e.g. if results of one method gave consistently tighter results than another, this might lead to potential reductions in MRLs, by knowing that such limits were routinely achievable). Other subjective factors that required consideration included laboratories' potential experience level in achieving tighter limits for a given method. If there was no previous requirement to report at low levels, labs might not have modified their calibration or QC procedures accordingly. Arsenic was used for one case study because it has highly documented health risks at very low levels, leading regulators to search for the lowest possible reporting limits, even at the risk of limiting the number of available labs. The other case study is for the element barium, for which the analytical accuracy and precision by any of the approved methods are well known to be adequate for addressing health risks. Thus, in this case, the regulatory approach should be based on maximizing the number of potential labs that can do the testing. For arsenic, if all the labs in the study were included along with all approved methods, the determined MRL would be on the order