Acute adrenal insufficiency has been demonstrated in a number of patients with shock. This study was designed to evaluate the rate of occult adrenal insufficiency in the critically ill trauma population and to determine the impact of hypoproteinemia on the use of random cortisol levels as a marker for adrenal insufficiency. Forty-four patients were prospectively enrolled on admission to the trauma intensive care unit, with three excluded, for a total n of 41. Random total serum cortisol and albumin levels were drawn on hospital Days 1, 4, 8, and 14. Occult adrenal insufficiency was defined as a cortisol less than 25 mcg/dL in the setting of an albumin greater than 2.5 g/dL. The prevalence of cortisol less than 25 mcg/dL ranged from 51 to 81 per cent during the study period, and peaked on Days 4 and 8. Albumin 2.5 g/dL or less ranged from 37 to 60 per cent, and this prevalence also peaked on Days 4 and 8. The patients with a low albumin had a high prevalence of low cortisol, ranging from 67 to 100 per cent. The prevalence of adrenal insufficiency, with low cortisol and normal albumin, ranged from 41 to 82 per cent during the study period. None of our patients with occult adrenal insufficiency were treated with steroids, which was a decision made by the treating physicians. Among the patients with occult adrenal insufficiency, survival was 100 per cent. Occult adrenal insufficiency is common in critically ill trauma patients, and is a dynamic entity that can be acquired and even resolved during critical illness. Random cortisol of 25 mcg/dL may actually not be an adequate marker of occult adrenal insufficiency. Low albumin predicts a low cortisol. Hemodynamically stable occult adrenal insufficiency should not be treated with steroid replacement in the critically ill trauma patient, as survival in our series was 100 per cent without replacement.