We aimed to review the proportion of influenza-like illness (ILI) that could be attributed to influenza infection over five consecutive influenza seasons in Victoria, to compare ILI activity with hospital admissions for influenza, and to develop thresholds that would indicate when ILI activity may coincide with increases in influenza hospitalizations. Combined nose/throat swabs from patients with ILI selected from sentinel general practices were tested for influenza and other respiratory viruses at the Victorian Infectious Diseases Reference Laboratory. The proportion of ILI attributed to laboratory-confirmed influenza was evaluated for five consecutive surveillance seasons, from 1998 to 2002. The seasonal patterns of ILI and laboratory-confirmed influenza were compared for 2000-02. ILI surveillance from sentinel general practices between 1997 and 2000 was compared with hospital admissions for influenza extracted from the Victorian Admitted Episodes Dataset between 1994 and 2000. Approximately 41% of all ILI cases from sentinel general practices were confirmed to have an influenza infection between 1998 and 2002. The seasonal pattern of ILI and confirmed influenza among patients selected from sentinel general practices was very similar over three influenza seasons. The trends for ILI rates in sentinel general practices appeared to correlate remarkably well with those of hospital admissions for influenza. Thresholds for ILI were defined as: baseline < 0.25 ILI cases/100 patients/week; normal seasonal activity 0.25-1.5; above normal seasonal activity > 1.5-3.5; and epidemic activity above 3.5. ILI rates from sentinel general practices correspond with hospital admissions coded as influenza and allow the use of threshold levels to describe seasonal influenza activity.