Recent work in behavioral economics has shown that stress increases choices of smaller, sooner gains relative to larger, later monetary gains. The simplest model that explains these findings is one in which stress increases the discount rate or present bias. A sharp test of this model is provided by intertemporal choices in the losses or effort domain: this model predicts that stress should lead to increased choice of larger, later losses or effortful tasks relative to smaller, earlier ones. Here we show suggestive evidence for the opposite result: using a laboratory experiment with 578 participants from informal settlements in Nairobi, Kenya, we find that stress increases choices of smaller, sooner outcomes across domains. Specifically, we show that the effect is present in monetary gains and losses, and effortful tasks; and for both a psychosocial stressor (Trier Social Stress Test), and the pharmacological elevation of stress hormone levels using hydrocortisone. Importantly, the results are statistically robust only in the absence of clustering, and should thus be regarded as tentative. However, they are at least initially consistent with a model in which stress increases discounting in the gains domain but decreases it in the losses and effort domains; or with a model in which stress decreases the utility of any future outcome. Thus, stress may affect intertemporal choice, but may do so through mechanisms other than a simple increase in discount rates or present bias.