In a trial, patients who came to a hospital Emergency Department (ED) with non-urgent complaints were advised and referred to primary health care outside the hospital. The effect of this was assessed by measuring health care utilization one year before and one year after the referral, using the Stockholm County computerized medical information system and ED medical records. The proportion of the 189 referred patients who visited the ED decreased from 48% to 42%, whereas in a control group of 107 patients the proportion increased from 41% to 51%. A small proportion, 7%, of the referred patients with four or more ED visits accounted for 45% of the total number of ED visits the year before the trial. These frequent ED users did not reduce their ED utilization more than frequent ED users in the control group. The use of health care centres increased in the referred group and was practically unchanged in the control group. However, those referred patients who continued to use the ED still quite often did so for non-urgent complaints.