BACKGROUND: Because total thyroid hormone testing is performed on many automated clinical chemistry instruments, the IFCC Scientific Division commissioned the Working Group for Standardization of Thyroid Function Tests to include total thyroxine (TT4) and total triiodothyronine (TT3) in its standardization efforts. METHODS: Existing SI-traceable reference measurement procedures (RMPs) were used to assign TT4 and TT3 values to 40 single-donor serum samples for subsequent use in a method comparison study with 11 TT4 and 12 TT3 immunoassays. Data from comparison of each immunoassay with the RMPs provided a basis for mathematical assay recalibration. RESULTS: Seven TT4 assays had a mean bias within 10% of the RMP, but 2 deviated by an average of -12% and another 2 by +17%. All TT3 assays showed positive biases, 4 within and 8 outside 10%, up to 32%. Mathematical recalibration effectively eliminated assay-specific biases, but sample-related effects remained, particularly for TT3. Correlation coefficients with the RMPs ranged from 0.82 to 0.97 for TT4 and from 0.32 to 0.92 for TT3. The within-run and total imprecision ranges for TT4 were 1.4% to 9.1% and 3.0% to 9.4%, respectively, and for TT3 2.1% to 7.8% and 2.8% to 12.7%, respectively. Approximately one-half of the assays matched the internal QC targets within approximately 5%; however, we observed within-run drifts/shifts. CONCLUSIONS: The study showed that of the assays we examined, only 4 TT4 but the majority of the TT3 assays needed establishment of calibration traceability to the existing RMPs. Most assays performed well, but some would benefit from improved precision, within-run stability, and between-run consistency.