A two-year follow-up study of treatment default among 193 patients with pulmonary TB, who were receiving domiciliary treatment in a Delhi urban clinic, revealed that the percentage of defaulting (that is, collecting drugs for less than 10 months) fell from 57% to 44% when the duration for calculating drug collection was raised from 12 to 24 months. The propensity to default appeared to be inversely related to the precision of diagnosis and the extent of lesions. While the default rate was 20.2% among those who were initially sputum positive, it was 100% among those sputum negative cases who had only minimal radiological lesions. This study, thus, questions the rationality of assessing the performance of a TB clinic on the basis of the ‘traditional’ definition of a defaulter. It has presented data to make a case for a more precise definition of a defaulter by offering a longer period for calculation of drug collection and by stressing the need for greater precision in diagnosis of cases who are put under treatment.