The response of Cd(Zn)Te Schottky and resistive detectors to intense x-rays is investigated in a commercial computed tomography (CT) system to assess their potential for medical diagnostics. To describe their signal height, responsivity, signal-to-noise ratio (SNR), and detective quantum efficiency the devices are modeled as solid-state ionization chambers with spatially varying electric field and charge collection efficiency. The thicknesses and pixel areas of the discrete detector elements are 0.5-2 mm and a few mm2, respectively. The incident spectrum extends from 26 to 120 keV and comprises 10(10) quanta/s cm2. It photogenerates a carrier concentration in the semiconductor that is two to three orders of magnitude above the intrinsic concentration, but remains to a similar extent below the charge densities on the device electrodes. Stable linear operation is achieved with the Schottky-type devices under high bias. Their behavior can be modeled well if negatively charged near-midgap bulk defects with a concentration of 10(11)-10(13) cm-3 are assumed. The bulk defects explain the amount and time constant (about 100 ms) of the detrapping current measured after x-ray pulses (afterglow). To avoid screening by the trapped space charge the bias voltage should exceed 100(V) x [detector thickness/mm]2. Dark currents are of the order of the generation-recombination current, i.e., 300 pA/mm3 detector volume. With proper device design the signal height approaches the theoretical maximum of 0.2 A/W. This high responsivity, however, is not exploited in CT since the SNR is determined here by the incident quantum noise. As a consequence of the detrapping current, the response speed does not meet CT requirements. A medium-term effort for crystal growth appears necessary to achieve the required reduction of the trap density by an order of magnitude. Scintillation based detectors are, therefore, still preferred in fast operating medical diagnostic systems.