Measuring depression in medically ill patients can be affected by the somatic symptoms occurring due to the underlying illness. This study used the Montgomery and Asberg Depression Rating Scale, over a 21-day period in 211 patients with acute life-threatening illness, to assess the influence of three factors upon the incidence of depression. These were time of measurement, the cut-off point used, and the effect of excluding somatic items. Using the standard rating scale, and a cut-off point for depression of 7, 63% of the patients were depressed on admission, 46% at one week and 37% at two weeks. Raising the cut-off point for depression to 14 and 21 led to 39% and 16% being depressed on admission and only 8% and 2% at one week respectively. In those patients scoring up to 14 over 75% of this score was accounted for by somatic symptoms. If the rating scale was modified to exclude somatic items from the score the rating scale performed more consistently, and the influence of somatic items was reduced. This study shows that the time of interview, the cut-off point used, and the influence of somatic items are all important in determining the incidence of psychiatric morbidity detected in this patient population.