In the first half of 1992, the Board of Directors at the Karolinska Hospital decided that, as of 1 September 1992, smoking indoors would be prohibited for staff in the hospital. A year later, a new Tobacco Act was introduced in Sweden involving a general ban on smoking in all health care premises, including rooms which patients would have cause to visit or frequent regularly. Four years after the implementation of the policy decision, an evaluation was made based on the principles of data and method triangulation. In all, four sub-studies were carried out: (1) an observational study about outdoor smoking; (2) interviews with gardeners, cleaners and hospital hostesses/hosts; (3) inventory of sales points for tobacco products, the extent of tobacco advertising and an interview with the staff at a shop and a café; and (4) an observational study about the appearance of signs on buildings about the smoking policy. The results of the study shows that smoking was found mainly among the staff of the hospital, particularly around the entrances. No visible tobacco advertising could be seen at the hospital sales point, and a drop in tobacco sales was noted. The experiences of a smoke-free Karolinska Hospital are that a policy decision requires continuous follow-up and evaluation. Furthermore, the actions taken must be subject to quality control, and, simultaneously, smoke-quitting activities must be organized for staff who still smoke.