We tested symptom-based surveillance during the G8 conference in 2000 as a means of detecting outbreaks, including bio-terrorism attacks, promptly. Five categories of symptoms (skin and haemorrhagic, respiratory, gastrointestinal, neurological and unexplained) were adopted for the case definition of the surveillance. The surveillance began 1 week before the conference, and continued until 1 week after the conference ended. We could not detect any outbreaks during this surveillance. Compared to the existing diagnosis-based surveillance system, symptom-based surveillance has the advantages of timeliness and simplicity. However, poor specificity and difficulties in determining epidemic threshold were important limitations of this system. To increase the specificity of surveillance, it is essential to incorporate rapid laboratory diagnoses into the system.