To study recurrence rate over a long period after recovery from previous tuberculosis history, we examined the frequency of previous tuberculosis history in patients who were admitted to our hospital in 1980-83 and in 1997-99 and the comparison was made between cases with and without culture-positive tuberculosis. The tuberculosis groups comprised of 297 patients in 1980-83 and 688 patients in 1997-99. The non-tuberculosis groups (control groups) comprised of 373 patients in 1980-83 and 1092 patients in 1997-99 with non-tuberculosis diseases other than the tuberculosis-related diseases such as non-tuberculosis mycobacteriosis, pulmonary aspergillosis, bronchiectasis, chronic bronchitis and tuberculosis sequelae. The patients with viral chronic hepatitis previously operated and transfused were also excluded as they might be operated because of pulmonary tuberculosis in the era of surgical treatment for tuberculosis. In both tuberculosis and control groups, they had previous tuberculosis history most frequently when they were twenties. In the control groups, the frequency of previous tuberculosis history among cases admitted in 1980-83 and were born in 1910-19, 20-29, 30-39, 40-49 were 15/84 (17.9%), 22/93 (23.7%), 11/77 (14.3%) and 3/43 (7.0%), respectively, and those admitted in 1997-99 were 11/70 (15.7%), 30/231 (13.0%), 28/288 (9.7%), and 10/230 (4.3%), respectively. In these 4 birth year groups, frequency of previous tuberculosis history among cases admitted in 1997-99 were significantly lower than that admitted in 1980-83 (p < 0.05, one-sided paired t-test), and the fact suggests that persons with tuberculosis history died earlier than those without it. In the tuberculosis groups, the frequencies of previous tuberculosis history among cases admitted in 1980-83 and were born in 1910-19, 20-29, 30-39 and 40-49 were 20/35 (57.1%), 31/58 (53.4%), 19/48 (39.6%), and 11/53 (20.8%), respectively, and those among cases admitted in 1997-99 were 30/99 (30.3%), 58/125 (46.4%), 22/102 (21.6%) and 17/136 (12.5%), respectively. The frequency of previous tuberculosis history among cases admitted in 1997-99 was significantly lower than that admitted in 1980-83 (p < 0.01) as was the case in the control groups. As recurrence within 5 years had occurred in only 4 out of 113 tuberculosis patients (3.5%) in the above-mentioned 4 birth year groups, almost all tuberculosis patients were assumed to have recovered completely from previous tuberculosis. Comparison between the recurrence rate from previous tuberculosis and the incidence rate from the remotely infected persons without previous tuberculosis history in the same birth year group can be done by calculating the prevalence of tuberculosis infection for each birth year group using a model of annual risk of tuberculosis infection appropriate for Japanese. The ratios between the recurrence rate from previous tuberculosis patients and the incidence rate from remotely infected persons without previous tuberculosis history were 4.71, 2.33, 1.78 and 1.11 in 1980-83 and 1.84, 3.99, 1.80 and 1.11 in 1997-99 for groups born in 1910-19, 20-29, 30-39 and 40-49, respectively. The ratio did not change systematically with time in these groups, indicating the recurrence rate did not change with time more than ten years after recovery from previous tuberculosis. The ratio was about 3 for groups born in 1910-19 and 20-29 and 1 for group born in 1940-49. Almost all patients born in 1940-49 could receive chemotherapy for tuberculosis in their twenties, while most of the patients born in 1910-29 could not. Therefore, the above-mentioned fact may reflect the recurrence rate of patients treated successfully with chemotherapy is almost the same as the incidence rate from remotely infected persons, while that the recurrence rate from previous tuberculosis patients spontaneously recovered is 3 times higher than the incidence rate from remotely infected persons.