In an effort to determine whether cues related to changes in energy contribute to the detection of a tone added to a narrow band of noise, we examined the effect of level variation on detection thresholds. In the first experiment, the level of each waveform was randomly varied on each presentation. Level variation had only marginal effects on performance. In addition, detection thresholds were obtained using bands of noise with equal energy across intervals. Neither increasing nor decreasing the variance of the noise-alone and tone-plus-noise energy difference distributions altered the detectability of a tone added to noise. Thus, the changes in energy that are concomitant with the addition of the tone are not the sole cue for the detection of the tone. In a second experiment, three psychometric functions were measured. One function was determined using no level variation, one was measured in the presence of level variation, and one was measured in the context of level variation, but for trials without level variation. The context of level variation slightly reduced detectability. In a third experiment, we compared detectability in three conditions: no level variation, across-trial level variation, and across-interval level variation. The thresholds obtained in the absence of level variation were superior to those measured in the presence of level variation, regardless of the manner in which the level variation was incorporated.