Thresholds for the detection of increments and decrements in level of 70 dB SPL sinusoidal signals were measured as a function signal duration (10, 20, or 200 ms), pedestal duration before the signal (10 ms, 200 ms, or pedestal on continuously) and frequency (250, 1000, or 4000 Hz). The sinusoids were presented in a low-pass filtered background noise with an overall level of 68-69 dB SPL which had two purposes: (1) to mask spectral splatter; (2) to induce an adaptation effect, which caused the continuous 4000-Hz pedestal (but not the other two pedestals) to decay to inaudibility (adaptation). We were particularly interested in determining whether the difference in noise-induced adaptation across frequency would influence the pattern of results. Seven normal-hearing subjects were used. Thresholds improved with increasing frequency and with increasing duration for both increments and decrements. However, the effect of increment/decrement duration decreased with increasing frequency; at 4000 Hz thresholds were almost the same for increment durations of 10 and 20 ms. The energy of the increments at threshold increased markedly with increasing increment duration (especially from 20 to 200 ms), suggesting a dominant role for the onsets of the increments as opposed to ongoing differences in level. Increasing the pedestal duration before the increment from 10 to 200 ms slightly improved thresholds for increment and decrement durations of 10 and 20 ms. Increment thresholds were similar for the gated and continuous pedestals at all frequencies, even though the 4000-Hz continuous pedestal decayed to inaudibility. However, thresholds for 200-ms increments were somewhat lower for continuous than for gated pedestals, and supplementary experiments found a larger gated-continuous difference for pedestals presented in quiet. Making the pedestal continuous adversely affected performance for the 10- and 20-ms decrements, but not for the 200-ms decrement. We suggest that the results for decrement detection may be affected by neural long-term adaptation, although they are not clearly related to loudness adaptation.