A problem in assessing animal welfare is that collecting data in itself may be stressful to the animals. Therefore, noninvasive methods for collecting data have to be devised and tested. A first step in investigating saliva cortisol, urinary cortisol, and urinary catecholamine as noninvasive indicators of canine well-being is the validation of these hormonal measures as alternatives for those in plasma. Using a model of insulin (0.2 U/kg)-induced hypoglycemia, we report on stress-induced responses in saliva cortisol, urinary cortisol, and urinary catacholamines relative to cortisol and catecholamine responses in plasma. Hypoglycemia in six dogs induced significant (P < 0.05) increases in plasma cortisol and adrenaline but not noradrenaline. Saliva cortisol responses expressed as net area under the response curve correlated significantly with plasma cortisol responses (r > 0.92). Saliva cortisol levels measured 7 to 12% of plasma cortisol concentrations. Cortisol/creatinine rations in urine were significantly higher when voided after insulin administeration, compared to when voided after saline treatment. Insulin-induced increments in cortisol/ creatinine ratios were nonsignificant when urine samples were assayed after dichloromethane extraction. Although urinary adrenaline/creatinine (A/C) ratios were significantly correlated with maximum plasma adrenaline values after insulin administration, A/C ratios did not differ significantly between insulin and saline treatment. The present experiment provides strong support for using saliva sampling and urine collection as noninvasive methods to establish stress-induced cortisol responses. For measuring acute plasma adrenaline responses, measuring A/C ratios may not be a valid alternative.