We reviewed reported pneumonia cases and deaths in Thailand since 1975 to evaluate the pneumonia surveillance system. In Sa Kaeo Province, we analyzed 3 years in detail (1999--2001) from electronic surveillance data, and compared deaths reported through surveillance to death certificate data in 1999 and 2000. In addition, we interviewed surveillance personnel who collected the data from all 7 hospitals and from a 10% random sample of health centers. Since the mid-1980s, reported illnesses and deaths from pneumonia have been increasing. In Sa Kaeo, an average of 925 pneumonia cases were reported each year, for an estimated average annual incidence of 211 per 100,000. The age-specific incidence peaked at 1,418 per 100,000 in children less than 5 years. In 1999 and 2000, there were 7 and 6 pneumonia deaths, respectively, reported through the surveillance system, compared with 28 and 53, respectively, reported by death certificate. Sixty-two (82%) of the 72 surveillance personnel reported receiving some training, but most of this was informal. Although written criteria to diagnose pneumonia were established in 1996, those who report cases did not know these criteria. A combination of physician, nurse, and public health workers diagnoses were used. According to the written criteria, cases of suspect or rule out pneumonia should be reported, but when asked about specific examples only 79% of persons interviewed said they would report "tuberculosis with pneumonia" and 44% would report "bronchitis, rule out pneumonia." Seventy-four percent of persons interviewed completed the surveillance report within one day of patient admission.