The standard model used by the International Commission on Radiological Protection (ICRP) to estimate the biological half-life of tritium in reference to man is an inadequate approximation. Both experimental data and a system's analysis of probable tritium kinetics indicate that tritium loss is neither constant nor follows a single exponential loss pattern but is the sum of two or more exponential losses. The ICRP model substantially underestimates the biological half-life and therefore, the hazard. The authors' models provide a more realistic approximation of tritium dynamics in the body and allow better visualization of tritium pathways and system response. An attempt has been made to model tritium carried bound in metabolizable food to the cell in order to study system response. The models predict half-lives approximately twice as great as the 12-day half-life derived from the ICRP model. Accordingly, it appears that the ICRP model underestimates the hazard of tritium exposure by a factor of two.