Individual cases of influenza are not reported nationally to the Center for Disease Control (CDC). Other methods of surveillance, therefore, must be used to estimate the extent and impact of influenza in the country. In the past, CDC has performed telephone surveys during the respiratory season and tabulated mortality due to pneumonia and influenza from 121 cities. The telephone data vary considerably from state to state and are only a general assessment of influenza activity. Tabulation of influenza-pneumonia deaths reflects more accurately the extent and impact of influenza but suffers from a 3-4 week-lag behind the actual clinical events. To improve influenza surveillance over the past 2 years, CDC obtained weekly numbers of emergency room visits to large community hospitals, school and industrial absenteeism, numbers of specimens submitted and numbers positive for influenza isolation from laboratories throughout the United States. Surveillance was most effective in large urban areas of the US where community hospital emergency rooms function as private physicians. Where people are more likely to consult a private physician than utilize a hospital emergency room, the correlation between private physician visits and influenza was good. Furthermore, school and industrial absenteeism are not very sensitive indicators of influenza-A activity; however, school absenteeism was a good index of influenza-B activity.