In patients treated by intravenous (i.v.) thrombolytic therapy for cerebral ischaemia, a very early neurological improvement (VENI) [National Institutes of Health Stroke Scale score (NIHSSS) 0, or early improvement ≥5 points], predicts a favourable outcome. VENI patients are usually excluded from trials testing complementary strategies, but a few of them have bad outcomes. To determine why VENI patients may have bad outcomes, we analysed the reasons for bad outcomes [modified Rankin Scale (mRS) score 2-6 at 3 months] in consecutive VENI patients. Of 365 consecutive patients with a pre-stroke mRS 0-1 (185 men, median age 69 years, median NIHSSS 12, median onset-to-needle time 147 min), 71 (19.5 %) had VENI. They were more likely to have had recent transient ischaemic attacks (OR 3.64, 95 % CI 1.08-12.27), higher baseline NIHSSS (median 14 vs. 11, p = 0.003) and shorter onset-to-needle times (median 135 min vs. 151, p = 0.01), and they were less likely to develop pneumonia (OR 0.27, 95 % CI 0.09-0.76) or malignant infarction (p = 0.045). In the 21 VENI patients (29.6 %) with a mRS 2-6 at 3-months, bad outcomes were due to the residual deficit in 14, secondary worsening of ischaemia in 4, intracranial haemorrhage in 2, and death from cancer in 1. One-third of VENI patients have bad outcomes, due to the residual neurological deficit in most of them. This finding suggests that VENI patients who still have a significant neurological deficit 1 h after thrombolysis should not be excluded from trials testing complementary strategies.