Little is known about the sensitivity of surveillance for tuberculosis after integration of formerly dedicated tuberculosis surveillance and control into the general health care system, an integration which took place in Finland in 1987. We compared routine laboratory notifications to the National Infectious Disease Register (NIDR) for Mycobacterium tuberculosis from January 1, 1995, to December 31, 1996, with data collected independently from all laboratories offering M. tuberculosis culture, and with data from patient records. 1059 culture-positive cases were found. The overall sensitivity of the NIDR was 93 % (984/1059). The positive predictive value of a culture-positive case in the NIDR to be a true culture-confirmed case was 99%. For the culture-confirmed cases in the NIDR, one or more physician notification forms had been submitted for 89%. A highly sensitive notification system for culture-positive tuberculosis can be achieved in an integrated national infectious disease surveillance system based on laboratory notification.