patients who initially complained of abdominal pain and who were subsequently found to have no surgical problem when evaluated in the SAU. There were no complaints of adverse outcome for all patients discharged from the SAU both during and after the study period. We did not readmit any patients from the subgroup in question for further abdominal complaints, nor do we have knowledge that such patients were readmitted to neighbouring hospitals. The second question relates to cost analysis. Given our limited resources we could not structure our study differently. The ideal would be to randomize patients prospectively to both old and new types of admission system simultaneously, and then to perform diligent follow-up on all patients. Such a venture was clearly not a realistic proposition. The details of cost savings in this study at our hospital are clearly set out in the paper; however, savings anticipated at other hospital Trusts may vary depending on local conditions and structure of the acute surgical services. Consultants and Trust accountants should note that, although a significant number of urgent GP referrals may be inappropriate, this figure may vary substantially from region to region.