Threshold signal‐to‐masker ratios were obtained or a 1000‐Hz signal presented with either an identical frequency sinusoidal masker or a complex masker consisting of the same frequency as the signal plus two sidetones 300 Hz on either side of the central component. Data considered include signal durations of 20, 400, and 1000 msec with masker durations either equal to that of the signal or continuous for the block up‐and‐down, two‐interval, forced‐choice threshold run. The parameters of primary interest were the level of the masker and whether the masker was gated or continuous. The thresholds appear to indicate that when the excitation pattern of the masker is increased (by shortening its duration or adding sidetones) so as to equal or surpass that of the signal, masked thresholds tend toward an inverted‐N function, as related to masker level, rather than being independent of masker level. Also, differences between the gated and continuous masker thresholds do not appear to be related to changes in signal uncertainty as reflected in the slopes of psychometric functions.