We used a computer-based method to help validate the reference ranges of assays for triiodothyronine (T3) and thyroxin (T4). A retrospective search of a database of laboratory results for the previous six months identified all patients with apparent euthyroid status, as defined by methods independent of the immunoassay under review. A computer-generated reference group (CGR Group) of 2001 records had a gaussian distribution of T4 values and a reference range (mean +/- 2 SD) of 56-161 nmol/L, compared with the supplier's suggested range for euthyroid subjects (58-148 nmol/L) and an in-house range of 60-144 nmol/L for a group of 97 normal subjects. A similar CGR Group of 1902 records gave a reference range for T3 of 0.7-2.1 nmol/L (manufacturer's range 0.8-2.8; normal subjects 0.8-2.2). An attempt to devise a reference range for thyrotropin failed when we found that its concentration in the population of patients with normal values for thyroid hormones was distributed differently from that in the normal population. The method is intended to be used in addition to conventionally derived ranges based on results for healthy subjects. It allows the laboratory to conveniently verify the reference ranges for T3 and T4 assays at regular intervals by using very large samples with appropriate age, sex, and weight distribution, drawn from the population of patients' samples submitted for analysis.