As part of a large nationwide case-control study covering a region with 78% of all births in New Zealand during 1987-90, the clothing and bedding of infants dying of sudden infant death syndrome (SIDS) and that of an appropriate control group were recorded. Cases consisted of 81% (n = 393) of all cases of SIDS in the study area and 88.4% (n = 1592) of 1800 control infants randomly selected from the hospital births and who completed a home interview. Use of a wool 'waterproof' underblanket was associated with a significantly reduced risk of SIDS (adjusted OR 0.44; 95% CI: 0.26-0.73) while sheepskin use was not. Firm tucking was identified as contributing to a reduced risk of SIDS even after adjusting for potentially confounding variables (adjusted OR 0.63, 95% CI: 0.46-0.86). Sixty case infants (15.6% of cases) were found dead with the head covered but there were no equivalent data for controls. Having been found previously completely covered by bedding was equally common in cases and controls (28.8% cases and 30.6% of control infants). Other differences of bedding and clothing between cases and controls were small; mattress characteristics were not studied. The exact methods in which babies are cared for are important and this study suggests that infants are at lower risk of SIDS when firmly tucked in and when sleeping on a 'waterproof' wool underblanket.