Hypothyroidism is a ubiquitous condition. According to the National Board of Health and Welfare, thyroid-hormone replacement was prescribed to approximately 370,000 individuals in Sweden during 2010. Eighty-three percent were women, representing more than 6 % of the female population.1 Autoimmune thyroiditis is the most predominant cause, but it may also be a consequence of previous radioiodine therapy or thyroidectomy. Hoarse voice, dry skin and muscle cramps are common symptoms. Intensified fatigue, coldness, tendency to constipation or depressive symptoms will often force the patient to a physician. An increased thyroid-stimulating hormone (TSH) – indicating a primary hypothyroid insufficiency – may contribute to the diagnosis of hypothyroidism. However, a weak association is reported between a higher TSH and a number of symptoms.2 A hypothyroid patient is commonly treated with levothyroxine and the dose adjusted until TSH is normalised.3 Despite treatment resulting in normalised serum TSH Saravanan et al. have reported that patients show a permanent impairment in psychological wellbeing compared with healthy subjects.4 A coupling to the lower triiodothyronine (T3) to thyroxine (T4) ratio noted in levothyroxine-treated patients compared with healthy subjects5–8 has been proposed to be a contributing cause. Thus, combined T3 and T4 treatment have been tried to improve patient’s wellbeing. However, a recent meta-analysis of 1,200 patients demonstrated ambiguous results when searching for clear correlation.9 Another study using higher T3 supplementation supported the notion that an increased T3 level may be associated with a better wellbeing.10 The conversion from T4 to T3 is regulated by deiodinase and the activity is elevated in the hypothyroid state as supported by in vitro studies in which thyroid cells incubated with TSH demonstrated increased deiodinase and messenger RNA (mRNA) activity.11,12 Our study’s aim was to determine if withdrawal of oral levothyroxine would elevate TSH and free T3 (fT3) to free T4 (fT4) ratios due to an increased conversion of fT4 to fT3. Bearing in mind the limited number of patients studied it was not intended to evaluate changes in symptoms. Materials and Methods Patients with primary hypothyroidism, 18–70 years of age, regularly controlled at our out-patient clinic, the Medical Department at Halland County Hospital, Sweden, were registered. The 71 patients who had treatment with levothyroxine alone without dose adjustments during the last three months, with normalised fT4 levels and normal or subnormal TSH were sent a written invitation. Patients with a symptomatic cardiac disease undergoing treatment with drugs that might affect the levels of thyroid hormone in the last three months were excluded (iodinecontaining X-ray contrast and antidepressants). In addition, pregnant and nursing women were also excluded. A total of 13 patients, after written consent was received, were included in our study During basal treatment TSH, fT3 and fT4 plasma levels were monitored on two consecutive days and a mean value was calculated (day 0). After obtaining baseline values, daily oral treatment was discontinued. Thereafter, plasma hormone level sampling was repeated after three, six and seven days before re-initiating ordinary treatment. The samples were obtained in fasting state at 8 am and analysed at the Department of Clinical Chemistry of Halland County Hospital, Halmstad. The assay used was an electro-chemical-luminescence immunoassay manufactured by Roche Diagnostics GmbH (Mannheim, Germany).13 The reference range for TSH was 0.40–4.00 mU/l, the inter- and intraassay coefficient of variation (CV) were 3.8 and 9 %. The reference range for fT3, 3.1–7.0 pmol/l, inter- and intra-assay CV were 5.5 and 7.4 %, respectively. The reference range for fT4, 11–22 pmol/l, inter-and intra-assay CV were 5.2 and 7.4%, respectively. The study was approved by the Regional Ethical Review Board in Lund 2011-02-17 DNR 2011/66. Consent was received from each patient after full explanation of the purpose and nature of all procedures used. Statistical methods: T3/T4 and TSH-levels across the observation points (baseline, days 3, 6 and 7) were compared using Friedman’s test.14 Moreover, paired levels between adjacent observation points (baseline versus day three, day three versus day six and day six versus day seven) were compared by Wilcoxon’s signed rank test; and the p-values reported were adjusted for multiple comparisons by the Bonferroni method.14 The methods used to calculate p-values were exact, rather than asymptotic.