Lung involvement accounts for significant morbidity and is a leading cause of mortality in patients with systemic sclerosis (SSc). It has been shown that different patterns of pulmonary involvement are seen in different subtypes of SSc. This paper reports a retrospective review of 72 patients with SSc to determine whether disease classification according to the extent of skin involvement alone (diffuse vs. limited) or autoantibody status was predictive of pulmonary parenchymal involvement. The diagnosis of interstitial lung disease was based on pulmonary function tests and chest radiographs. Restrictive lung disease was common in both limited SSc (lSSc) and diffuse SSc (dSSc), occurring in 30% and 50% of these patients respectively (P = 0.16). Radiographic evidence of significant interstitial disease was also comparable between the groups [nine of 32 lSSc patients (28%) vs. six of 17 dSSc patients (32%), P = n.s.]. No significant difference in mean lung function was found between patients with anti-Scl 70 antibody (n = 12) compared to those without (n = 60) (TLC 79.0 +/- SE 5.1% predicted vs. 82.8 +/- 2.2, P = n.s.; DLCO 63.0 +/- 5.1 vs. 59.7 +/- 2.5, P = n.s.). By contrast, statistically significant differences in mean lung function were found between patients with anticentromere antibody (ACA) (n = 24) and those without ACA (n = 48) (TLC 98.6 +/- SE 3.9% predicted vs. 79.7 +/- 3.1%, P < 0.001); and less frequent radiographic evidence of severe interstitial disease (0 of 17 with significant interstitial changes on chest radiograph vs. 15 of 32 (47%), P = 0.002). It is concluded that classification of SSc patients on the basis of the distribution of skin involvement poorly predicts the occurrence of interstitial lung disease. On the other hand, ACA is highly associated with the absence of interstitial lung disease.