Granuloma formation is seen in a variety of pathological settings. These structures are required for protection against infectious organisms such as mycobacteria but can also be detrimental to the host, because the organs harbouring the granulomas undergo adjacent tissue destruction and displacement due to their presence (reviewed by Agostini & Semenzato 2003; Kaye & Engwerda 2003; Schwander & Ellner 2003; Turner et al. 2003). Typically, granulomas are defined as focal accumulations of mononuclear cells in various stages of differentiation that generally appear as well-structured lesions consisting of centrally located macrophages which are surrounded by a lymphocytic cuff, and by an exterior fibrotic capsule. This encapsulation serves as a demarcation of the granulomas from the surrounding tissue (Benini et al. 1999; Florido et al. 2002) and physically restrains mycobacterial spread (Saunders et al. 1999; Ehlers et al. 2000; Saunders & Cooper 2000). In a pro-fibrotic response, the arginase present in macrophages is activated and leads to the production of proline and collagen (Hesse et al. 2001; Wynn 2003, 2004). This reaction has l-arginine as its substrate, the same substrate used by the inducible nitric oxide (NO) synthase (iNOS) for the production of NO. Therefore, it is not surprising that iNOS and NO production have been correlated with an antifibrotic effect in granulomatous diseases (Brunet et al. 1999; Hesse et al. 2000, 2001; Nascimento et al. 2002; Pearce & MacDonald 2002; Wynn 2004). The regulation of fibrosis by NO in mycobacterial granulomas has not been thoroughly studied. In a nonfibrotic granuloma model induced by purified protein derivative (PPD)-beads, NO was involved in the formation of the granulomas, and NO inhibition induced collagen deposition (Hogaboam et al. 1997, 1998). In human tuberculosis, active iNOS was detected in epithelioid and multinucleated giant cells in granulomas, and evidence was presented correlating iNOS expression and granuloma formation, independently of their aetiology (Facchetti et al. 1999). Fibrosis was studied in experimental tuberculosis using BALB/c mice and it was shown to be regulated by interleukin (IL)-4(Hernandez-Pando et al. 2004). In a different model involving the transfer of polarized T-helper cell subsets, it was shown that whereas Th1 lymphocytes conferred protection, Th2 cells were responsible for pulmonary fibrosis and increased morbidity during a high-dose Mycobacterium tuberculosis infection (Wangoo et al. 2001). The study of M. tuberculosis infection in iNOS-deficient mice to analyse the issue of fibrosis regulation by NO is problematic, because these animals are more susceptible to the infection than normal mice, and the correlation between lack of NO and fibrosis would be confounded by the differences in bacterial loads. Whereas in the case of infection by M. tuberculosis iNOS expression has been associated with protection (MacMicking et al. 1997), the same does not hold true for Mycobacterium avium infection (Doherty & Sher 1997; Ehlers et al. 1999; Florido et al. 1999, 2005; Gomes et al. 1999). Thus, the M. avium-infection model is better suited for an analysis of the role of NO in fibrosis induced by mycobacterial infection. iNOS expression was associated with granuloma formation in murine M. avium infection. Treatment with L-NIL, a selective iNOS inhibitor, significantly increased the number, size and cellularity of granulomas (Ehlers et al. 1999). iNOS-deficient (iNOS-KO) mice also had more and bigger granulomas when compared with wild-type (WT) mice, and those granulomas had a well-structured core of epithelioid macrophages, surrounded by a mantle of lymphoid cells (Gomes et al. 1999). In the current study, collagen deposition was analysed in granulomas of iNOS-KO mice infected with two M. avium strains of different virulence, and compared with WT mice. Although tissue deposition of collagen can be quantified by chemical detection of hydroxyproline, this type of analysis does not discriminate between deposition of collagen in the granulomas themselves vs. a generalized organ fibrosis often seen in infected hosts. Therefore, an image analysis program was designed, so that collagen content could be determined within the granulomas. The infection was via the intravenous route, in which case, most of the inoculum is trapped in the liver, and the granulomas formed are localized especially in that organ. Furthermore, the fact that granulomas and the surrounding parenchyma are clearly distinguishable in the liver justifies the choice of doing the analysis of those structures in the liver and not, for example, in the spleen where the borders of the granuloma are less clearly identified. We found that collagen content was higher in granulomas from iNOS-KO mice than WT mice when either strain was being used, that the collagen deposition increased as infection progressed and did not confine itself to delimitating the granuloma but also appeared deep within the granulomatous structure, and that the collagen deposition was noticeably lower in the case of an infection with the highly virulent strain.