When vertebrate physiological ecologists use the terms 'stress' or 'physiological stress', they typically mean the level of hypothalamus-pituitary-adrenal (HPA-) axis activation. Measurements of stress hormone concentrations (e.g. glucocorticoids in blood, urine or faeces), leukocytes (e.g. the neutrophil-lymphocyte ratio or heterophil equivalent), immunofunction (e.g. innate, cell-mediated or humoral immunity measures) and regenerative anaemia (e.g. mean erythrocyte volume and red blood cell distribution width) have all been used to estimate HPA-axis activity in free-living vertebrates. Stress metrics have provided insights into aspects of autecology or population regulation that could not have been easily obtained using other indices of population wellbeing, such as body condition or relative abundance. However, short- and long-term stress (often problematically termed acute and chronic stress, respectively) can interact in unpredictable ways. When animals experience trapping and handling stress before blood, faeces and/or urine is sampled, the interaction of short- and long-term stress can confound interpretation of the data, a fact not always acknowledged in studies of stress in free-living vertebrates. This review examines how stress metrics can be confounded when estimates of HPA-axis activation are collected for free-living vertebrates and outlines some approaches that can be used to help circumvent the influence of potentially confounding factors.