FDA Guidance for testing bioequivalence of levothyroxine (L-T(4)) preparations has been challenged by several groups, based on multiple issues. The efficacy of single versus combined hormone therapy also is receiving additional scrutiny. To examine these concerns, we developed a new nonlinear feedback system simulation model of whole-body regulation mechanisms involving dynamics of T(3), T(4), TSH, plasma protein binding, extravascular regulatory enzyme systems, and the hypothalamic-pituitary-thyroid axis, all quantified from human data. To address bioequivalence, we explored how to best account for varying and unmeasured endogenous T(4) following dosing with exogenous oral L-T(4) in euthyroid volunteers in required pharmacokinetic (PK) studies, by simulating various dosing scenarios and developing a new and simple correction method. We computed and assessed dosing error effects and baseline corrections using simulator-predicted endogenous T(4) level variations, due to actual closed-loop effects, and compared these with approximate corrections computed directly from PK data. Predicted dose-responses were quite linear, and for constant baseline, 7-day half-life, and our new formula-correction methods, we established some bounds on bioequivalent dosages. Simulated replacement after thyroidectomy required 141 microg L-T(4) only to normalize T(3) tissue levels and 162 microg L-T(4) to normalize plasma T(3) levels. A combined dose of approximately 103 microg L-T(4) plus approximately 6 microg T(3) ( approximately 18:1 ratio) was needed to normalize both plasma T(3) and T(4) and average tissue T(3) levels. However, simulated average tissue T(3) levels were normalized with standard L-T(4)-only therapy, and plasma T(3) levels were still within the normal range. We suggest a simple and more accurate correction for endogenous T(4) in PK studies. Current standard L-T(4)-only treatment is supported for routine replacement needs.