Serum free 3,5,3′-triiodothyronine (T3) levels are regularly reported reduced in patients with non-thyroidal, somatic illnesses (NTI). However most free T3 assays have serious methodologically shortcomings. From a theoretical point of view, ultrafiltration may be the most reliable technique at present, and we have previously reported unaltered serum free T3 levels in NTI. A newly commercialized assay suitable for routine use (Amerlite MAB free T3) has demonstrated promising results in NTI sera, and we thus compared these two methods in a large group of NTI patients not treated with glucocorticoids or dopamine (n = 120), as well as in 66 healthy controls and 8 patients on prolonged dopamine infusion. In both assays free T3 levels were unaltered in NTI (NTI versus controls (mean), Amerlite MAB free T3: 5.47 versus 5.32 pmol/l; ultrafiltration free T3: 6.99 versus 7.60 pmol/l). Free T3 levels outside normal range were found in 14% using the Amerlite MAB free T3 assay and 2% using the ultrafiltration assay. The two methods correlated in all subjects (n = 194) (r = 0.32(P < 0.001)) as well as in NTI patients (r = 0.34 (P < 0.001)). No correlations to serum T4 levels were found, and patients with serum T4 levels below or above 70 nmol/l, had similar free T3 concentrations in both assays. Patients on dopamine infusion had reduced serum free T3 levels (both assays), but the values were still within the normal range. Conclusion: Serum free T3 levels were unaltered in NTI patients using both assays, and the Amerlite MAB free T3 assay seems useable during diagnostic work-up for possible thyroid disease in hospitalized patients with other somatic illnesses, despite the fact that this assay has some shortcomings, which are an integral part of all immunoextraction assays.