OBJECTIVE: To describe the results of two Dutch experiments aimed at limiting the number of self referrals to accident and emergency (A&E) departments of a newly opened hospital. METHODS: Basic design for both field experiments was a one group test-retest study, with the opening of a new hospital being the experimental factor. Data refer (1) to the number of patient contacts for acute somatic and non-somatic professional help with traumata and intoxications filled in on contact registration forms, and (2) to patient reports based on written questionnaires and interviews by telephone. RESULTS: Both studies show that it is possible to limit the number of self referrals, although it is important to be aware of possible side effects. The solution chosen in one of the cities, with patients visiting the A&E department being sent back to a general practitioner, was highly effective but resulted in a series of complaints from the public. This forced the authorities to abandon their original policy. The model chosen in the other city, with two specialized health centres serving as an alternative for the free A&E department of the new hospital, was somewhat less successful in preventing patients from going directly to a hospital, but was much more acceptable for the public. CONCLUSIONS: Both studies showed that controlling the number of self referrals to the A&E departments of hospitals is possible. However, the more rigid the model, the more difficult it will be to have this model accepted by patients. Based on arguments like cost-effectiveness, a better solution might be not to discourage people from going to the hospital with minor injuries, but to integrate primary health care (PHC) and hospital facilities.