Three experiments were conducted to investigate the detectability of tonal signals simultaneously presented with computer-generated, long duration (.5 sec), band-reject maskers. In Experiment 1, detectability of a tone of 500, 1000, or 2000 Hz presented at 50 dB SPL was measured as a function of the width of a spectral "notch" symmetrically placed around the tone. A narrow notch resulted in decreased detectability relative to the wide band (zero-notch width) control condition. Further increases in notch width resulted in increased detectability until subjects approached errorless performance. In Experiment 2, psychometric functions were obtained for a 1000-Hz tone presented in four notch-noise conditions. The slopes of the psychometric functions were consistently steeper in the band-reject conditions relative to the zero notch-width condition. These slope differences result in relationships between threshold and notch width that depend upon the region of the psychometric function from which the threshold estimates are obtained. In Experiment 3, the decrease in performance associated with the notch was found to increase in proportion to notch depth. The data suggest that spectral edges placed in the immediate vicinity of a tone reduce the detectability of the tonal signal. This "edge effect" is discussed in relation to two-tone inhibition that has been observed in the auditory nerve.