Data are of unknown quality when the data user lacks critical information needed to guide interpretation of the data with respect to the intended decision. These data do not even rise to the level of "screening quality data" (which requires that the quality of data be known, i.e., individual performance parameters such has reporting limits, precision, bias, representativeness, etc.). This unfortunate situation sometimes occurs when data (field-generated or fixed-laboratory) are reported with no supporting QC and procedural information. There is no way for the data user to know whether proper sampling and analytical procedures were followed or not, or whether site-specific interferences caused bias or not. For example, data may be reported to the data user as simply a table of numbers or non-detects with no documentation about sample collection and preparation procedures, analytical method procedures, sample-specific detection limits, analytical precision, analytical bias, evaluation of interferences, locational information, etc. This is unacceptable and the data are unusable. The data user should never be asked to accept data results on "faith" and is fully justified in rejecting such data.

Legacy data sets are often common examples of data of unknown quality when pertinent QA/QC information either was never reported, or is no longer available. While likely unusable for decision-making purposes without further evaluation, these data can potentially be used in a qualitative fashion to support CSM development. For example, such data sets may identify contaminants of concern or give an indication of the magnitude of concentrations present. In some cases the value of legacy data sets can be verified by re-analyzing a subset of archived samples (if available) to provide points of comparison with data sets of known quality.