Two experiments using male rats evaluated the effects of a range of doses of the neurosteroid, pregnenolone sulfate (PS), or of the synthetic neurosteroid, ethylestrenol (E), on the retention of a passive avoidance task. The steroids either were given immediately after the training trial or 1 h before the first retention test. Retention tests were given both 24 h and 48 h after acquisition. In both experiments, separate groups of animals were trained under low or moderate footshock conditions. At all doses tested both PS and E improved retention under the low footshock conditions. In groups trained with the higher footshock, the steroid-treated groups performed no better than the vehicle controls. Indeed, there were suggestions that some doses impaired retention. These results seem best understood as an induction of bimodality or 'turbulence' in behavior as used in Chaos theory rather than a shift in an inverted U-shaped retention function. In the second experiment in which the steroids were given before retention testing, they were generally without effect.