Woo et al1 conclude that serum total cholesterol (TC) does not differ in Hong Kong stroke cases versus controls (5.3 versus 5.5 mmol/1, respectively). They further state that the only lipoprotein fraction which differed between intracerebral hemorrhage cases and controls was Lp (a)(higher in cases), and that patients with lacunar infarction had lower HDL3-C and Apo AI concentrations than controls. They base these findings on lipid and lipoprotein measurements made 3 months after ictus, 2 which forces them to restrict their analysis to 3-month survivors. We question their logic. Although they find2 that TC levels in stroke patients are lower 3 months after ictus than< 48 hours after ictus (5.3 mmol/1 versus 5.7 mmol/1, respectively), they also find that fatal strokes have lower TC than survivors (5.2 mmol/1 versus 5.8 mmol/1). More than half of their 81 fatal strokes were diagnosed with lacunar infarction (n= 8) or intracerebral hemorrhage (n= 37). We3 have found TC to be lower in fatal cases with intracerebral hemorrhage, as has the Honolulu Heart Program4 for incident cases with intracerebral hemorrhage. Woo et al2 state that the mechanism for lowered TC level 3 months after stroke is unclear. We raise the possibility that these levels at the time of stroke may be more representative of usual TC levels because poor nutrition or newly developed liver or renal dysfunction poststroke may result in lower TC levels 3 months later. At minimum, the admission TC level has the advantage of including fatal as well as nonfatal strokes. It appears that use of TC levels< 48 hours after ictus, including fatal strokes, will yield a substantially lower TC level in intracerebral hemorrhage cases than in controls in the data of Woo et al. We wonder if Dr. Woo would be kind enough to provide these data.