OBJECTIVE Two important characteristics for interstitial microwave antennas used in clinical hyperthermia are: (1) a good impedance match to minimize reflected power; and (2) a good power deposition pattern which is independent of insertion depth. A major problem of the miniature coaxial dipole antennas used for interstitial hyperthermia is the fact that the impedance and power deposition patterns of these antennas change with insertion depth. One possible solution is the addition of a coaxial choke. A theoretical model for calculating the input impedance of interstitial microwave antennas having a coaxial choke is presented, which may serve as the first step in the design of such antennas. METHODS A theoretical model for calculating the input impedance of coaxial microwave antennas with and without a choke is presented using insulated antenna theory. The theoretical model was used to calculate the input impedance of several prototype antennas having various choke and feedline dimensions, and comparison was made with experimentally measured impedance measurements in tissue-equivalent phantom. RESULTS The choke section of the antenna is not ideal if conventional plastic insulation is used as the choke dielectric, because the desired radiating length of the antenna is significantly shorter than the quarter-wavelength in the choke dielectric. Impedance calculations based on the theoretical model correlate reasonably well with experimentally measured impedance. Based on these calculations, the effect of parameters such as choke layer thickness and choke dielectric constant are discussed for a 915 MHz antenna with choke. CONCLUSIONS The theoretical model can serve as a design aid for optimizing choked microwave antenna designs, as well as predicting the impedance match of a given antenna design at a given insertion depth. The model allows the effect of some variables not accessible experimentally such as termination impedance to be studied, which may also be useful in the understanding of these antennas. Calculations are easily performed on a desktop computer.