Hyperglycaemia in acute ischaemic stroke is traditionally associated with a worsened outcome. However, it is unclear whether the impact of hyperglycaemia on stroke outcome is similar in lacunar and non-lacunar infarctions. The relation between serum glucose measured within 6 h after stroke onset and functional outcome was investigated in 1375 ischaemic stroke patients who had been included in two placebo-controlled trials with lubeluzole. The endpoint was a favourable outcome, defined as a modified Rankin Scale score < or =2 at 3 months. Classification into lacunar (n = 168) and non-lacunar (n = 1207) strokes was based on clinical criteria according to the Oxfordshire Community Stroke Project and findings on brain CT scan. Hyperglycaemia was defined as blood glucose >8 mmol/l. A possible concentration-dependent effect of glucose on outcome was investigated in both lacunar and non-lacunar stroke. Multivariate analysis showed that hyperglycaemia was associated with decreased odds of a favourable outcome in non-lacunar stroke (OR 0.60; 95% CI 0.41-0.88, P = 0.009), but with increased odds of a favourable outcome in lacunar stroke (multivariate OR for glucose >8 mmol/l: 2.70; 95% CI 1.01-7.13, P = 0.048). In non-lacunar stroke, there appeared to be a concentration-effect relation, as the odds of favourable outcome gradually decreased with increasing glucose levels. In lacunar stroke, an association with favourable outcome was observed with glucose levels >8 mmol/l, but this beneficial effect diminished with more severe hyperglycaemia >12 mmol/l. In conclusion, hyperglycaemia has a detrimental effect in non-lacunar stroke, but moderate hyperglycaemia may be beneficial in lacunar stroke.