In 1993 an epidemic caused by dengue virus type 2 occurred in several North Queensland population centres. Charters Towers, estimated population 10000, had 155 officially notified cases. An analysis of symptoms was undertaken using a random sample of 1000 residents to determine specificity of symptoms, the subclinical infection rate, and to establish the true extent of the epidemic. Retrospective diagnoses of dengue fever were based on the presence of both serum dengue 2 neutralizing antibody and presence of symptoms. An estimated 20% of the population had dengue fever. The rate of subclinical infections in this epidemic was 14·6%. There were no symptoms that were specific for dengue fever. Bleeding occurred more frequently in people who recalled a previous dengue infection during a dengue 1 epidemic 12 years earlier (55·6% vs. 16·8%, P=0·003). Surveillance for future epidemics should be based on serological and virological confirmation of dengue virus infection amongst symptomatic patient.