Two cyclic linear compartment models are proposed to investigate the reabsorption mechanism of doxycycline. In one model, reabsorption is considered to be continuous; in the other model, it is discontinuous. The continuous model, when fitted, leads to one real and two complex conjugate eigenvalues, corresponding to a regression equation consisting of a regular exponential term and an exponentially damped trigonometric expression. In spite of the apparent oscillatory nature of this regression equation, the fitted curves show no secondary peaks or humps apparent in the data. Simulation studies indicate that it may not be possible to get response profiles showing secondary peaks or humps that are experimentally detectable with linear compartment systems with cyclic pathways and continuous transfer. The model with discontinuous cyclic transfer was more flexible in describing the discrepancies in the data and appeared to be preferable to the continuous cyclic transfer model judged by the Akaike information criterion.