In this article, we discuss a novel signal processing technique for adaptive radar that permits joint target-matched illumination and interference avoidance in dynamic spectral environments. This approach allows for spectral coexistence between a radar system and a primary user of the radio frequency space. Spectral coexistence is exploited to allow use of higher bandwidths than would otherwise be available to conventional radar systems. The technique proposed exploits the relative simplicity of the error reduction algorithm, and also provides a novel use of the masking procedure to allow for target-matched illumination. Practical constraints such as constant modulus are considered in the waveform design procedure, while providing an implied signal-to-interference-plus-noise ratio improvement via the error reduction algorithm. Results for full simulation and hardware-in-the-loop experiments are presented and analyzed. We are able to show a signal-to-interference-plus-noise ratio gain of 40 dB is achieved for the target-matched waveform as compared with a linear frequency modulated waveform. However, the signal-to-interference-plus-noise ratio gain comes at a cost of degraded autocorrelation characteristics of the targetmatched illumination waveform, despite only modest levels of primary user spectrum occupancy. Spectral notch depths achieved by the modified error reduction algorithm are approximately 25 dB.