These experiments were designed to explore the benefit to signal detection of different types of across‐channel cues, both alone and in combination. Some conditions were similar to those used in profile analysis (PA), and some to those used in comodulation masking release (CMR). Others were designed specifically to eliminate, or render unreliable, a particular across‐channel cue so that the benefit to performance from another cue could be assessed. Thresholds for detecting an increment in level of a sinusoid, or of the carrier of a sinusoidally amplitude modulated (SAM) sinusoid, were measured in the presence or absence of four sinusoids or SAM sinusoids (flankers), two centered above and two centered below the signal frequency. The flankers were always modulated with the same depth as the target component during nonsignal intervals. The flankers, when present, were either equal in level to the nonsignal target sinusoid, or were scrambled in level (different in level both from each other and from the target by an amount that varied randomly from one stimulus to the next). In some conditions the overall level of the stimuli was also varied randomly from one stimulus to the next. The results indicate that about 5–6 dB of benefit arises from the cue of a disparity in level across frequency (a PA‐type cue), and about 1–3 dB from the cue of a disparity in envelope modulation depth across frequency (a CMR‐type cue). For some subjects, slightly less benefit occurred when the flankers were presented to the opposite ear as the signal, requiring across‐ear comparisons. Scrambling the level of the flankers often impaired performance, especially when the overall level of the stimuli was fixed. This appears to reflect an across‐channel interference effect.