An experiment was carried out to test the hypothesis that environmental treatments thought to change arousal in opposite directions (noise and loss of sleep) will also produce opposing changes in the degree of selectivity in attention allocation. A dual task method was employed, subjects having to perform a primary tracking task while simultaneously monitoring an array of light sources for occasional signals (the subsidiary task). The effect of loss of sleep was significantly to impair performance of the primary task, and to reduce the advantage of high-probability sources over low-probability sources, within the subsidiary task (i.e. a loss of selectivity). This result complements that found in previous experiments with noise (Hockey, 1970a, b), in which changes in attention allocation move in the opposite direction (an increase in selectivity), and thus supports the view that selectivity is a function of arousal level.