In-hospital delays to thrombolysis were significantly shorter when thrombolysis was available on admission to the accident and emergency department than after transfer to the coronary care unit (median 60 min v 84 min, p < 0.0001). With direct admission by general practitioners to a coronary care unit the subsequent in-hospital delay was shorter (median 39 min p = 0.0004), but overall delay to thrombolysis longer (median 220 v 170 min, p = 0.0019) because of longer pre-hospital delays. Overall delay was shortest with emergency ambulance referral and thrombolysis being administered in the accident and emergency department.