Adaptive interference cancellation algorithm using speech mode dependent thresholds. A method of processing radio frequency (RF) bursts dependent on a speech mode associated with data contained within the RF burst is presented. Different voice modes, full rate, half rate, and adaptive multi-channel rates each may require different signal to noise ratio (SNR) conditions in order to be successfully processed. To improve the equalization, the SNR associated with the burst is estimated. Based on the SNR or other related conditions, a decision can be made as to whether or not an interference cancellation burst process should be implemented. For example, any one or more of SNR of the signal, a measure of colored noise within the signal, an indication whether the signal being noise limited or interference limited, and a channel profile of the signal may indicate the presence of interference requiring the cancellation of such interference.