Aims We undertook this study to analyse the pattern of childhood cutaneous tuberculosis prevailing in northern India over the past 25 years and to highlight differences from and similarities to adult tuberculosis. Materials and methods Clinical records of children with cutaneous tuberculosis who attended the Nehru Hospital attached to the Postgraduate Institute of Medical Education and Research, Chandigarh, India, between January 1975 to December 1999 were analysed. Results Four hundred and two patients with cutaneous tuberculosis were examined during the 25-year period of observation. These patients formed 0.1% of the total dermatology outpatients. Seventy-five (18.7%) of these 402 cases were children (≤ 16 years). There were 32 (42.7%) boys and 43 (57.3%) girls with a boy : girl ratio of 0.74 : 1. The majority of the children, 41 (54.7%) were in the 10–14 years age group. There was no significant boy or girl preponderance in any group other than in scrofuloderma (SFD), where girls significantly outnumbered boys at all ages. Of the various patterns of cutaneous tuberculosis seen, 40 (53.3%) had SFD, 30 (40.0%) had lupus vulgaris (LV), 3 (4.0%) had tuberculosis verrucosa cutis (TVC), 1 (1.3%) child each had tuberculids and tubercular gumma. No child had a tubercular chancre or acute miliary cutaneous tuberculosis. The neck was the commonest site afflicted with SFD and the face was the commonest site affected with LV. No child had generalized lymphadenopathy. Eighteen (60.0%) of the 30 children with LV had regional lymphadenopathy of which 15 (83.3%) had localized disease and 3 (16.7%) had disseminated disease. Of the 16 children with systemic organ involvement, 12 (75.0%) had regional lymphadenopathy. Of the 62 children in whom the data regarding vaccination status was available, 31 (50.0%) had been vaccinated and 31 (50.0%) had not. Among the vaccinated group no child had disseminated disease. Three (9.7%) children in the nonvaccinated group had disseminated disease. Information regarding Mantoux reactivity was available in 71 (94.7%) children, 61 (86.0%) with localized disease and 10 (14.1%) with disseminated disease. Of the 61 children with localized disease, 56 (91.8%) were Mantoux positive and of the 10 children with disseminated disease, only 5 (50.0%) were Mantoux positive (> 10 mm). Histopathologic reports were available for evaluation in all 75 children. Out of 30 cases of LV, 24 (80%) showed classical tuberculous histology. In contrast, out of 40 cases with SFD, only 19 (47.5%) showed classical histology. Classical tuberculous histology was noted in all 3 (100%) cases of TVC and 1 (100%) case each with tuberculids and gumma. Tubercle bacilli could be demonstrated in 4/30 (13.3%) cases with LV and 9/40 (22.5%) cases with SFD. Systemic involvement was seen in 16 (21.3%) children, of whom 3 (18.8%) had LV and 13 (81.3%) SFD. The lungs were the most common organs involved in 8 (50.0%) children followed by bone(s) in 4 (25.0%), abdomen in 2 (12.5%), and both lung and bone in 1 (6.3%) child. In contrast to adults, girls outnumbered boys in the childhood series; SFD was a common form of presentation in contrast to LV and TVC, tuberculous gumma and tuberculids were noted less often. In both children and adults, Mantoux reactivity did not correlate with the extent of the disease; patients with disseminated disease were found to be less often vaccinated with BCG and regional lymphadenopathy was noted more often in patients with disseminated disease. Conclusions In the whole spectrum of cutaneous tuberculosis, there is a proportion of patients with dissemination (systemic involvement) who are of great epidemiological significance as they require a change in the standard therapeutic regimens recommended for cutaneous tuberculosis.