Disclosed herein are various mechanisms for the adaptive cancellation of radio frequency interference (RFI) in Discrete Multitone (DMT)-based signal transmissions systems. various mechanisms for the adaptive cancellation of radio frequency interference (RFI) in Discrete Multitone (DMT) modulated signal transmissions systems. In at least one embodiment of the present invention, RFI introduced by one or more RFI disturbers (e.g., ham radios) into a DMT modulated signal may be detected by calculating an RFI canceller coefficient for some or all of the DMT bins of the signal, where the RFI canceller coefficient represents the correlation between the noise at the corrupted bin and the reference RFI noise. The RFI canceller coefficients for the corrupted bins may be calculated using an LMS adaptation process. Those bins where the magnitude of the corresponding RFI canceller coefficient is greater than a certain threshold may be marked as corrupted by RFI. Further, one or more bins having RFI canceller coefficients with the largest magnitudes may be marked as reference disturber bins. The noise (e.g., error) present in the identified reference disturber bins may be determined using, for example, the slicer error. The RFI disturbance at some or all of the corrupted bins then may be estimated based at least in part on the error of the reference disturber bin proximal to the corrupted bin and the RFI canceller coefficient for the corrupted bin. The received signal then may be adjusted based on the RFI estimations for some or all of the corrupted bins to reduce or eliminate the RFI disturbance in the signal.