Cognitive performance, even during simple tasks, is impaired during acute hypoglycaemia (plasma glucose [PG] <2.5 mmol/l [45.0 mg/dl]) in individuals with type 1 diabetes mellitus [
1,
2] and the degree of impairment depends on the level of hypoglycaemia [
3]. Additionally, the cognitive impairment induced by hypoglycaemia (PG 2.5–2.7 mmol/l) can remain following the restoration of euglycaemia [
4‐
6]; with some studies showing cognitive impairments lasting for approximately 45–75 min after euglycaemic restoration [
7,
8]. Working memory is an important aspect of cognitive function and is susceptible to the effects of hypoglycaemia. It has been demonstrated using functional MRI (fMRI) that, compared with healthy individuals, people with type 1 diabetes mellitus require higher levels of brain activation to attain parity for working memory performance during hypoglycaemia (PG ≤2.8 mmol/l [50.4 mg/dl]) [
9]. Thus, identification of cerebral activation patterns during working memory performance at different PG levels could enhance our understanding of mechanisms underlying the reduced cerebral efficiency seen in type 1 diabetes mellitus [
9]. For example, it remains to be clarified if milder hypoglycaemic episodes, PG just below 3.1 mmol/l and previously associated with altered brain activity [
10,
11], have a similar negative impact on working memory to those demonstrated at PG ≤2.8 mmol/l. In individuals with type 1 diabetes mellitus, cerebral blood flow (CBF) and cerebral glucose metabolism [
12] within total grey matter are correlated. This correlation becomes stronger when adjusted for glucose levels, allowing CBF assessment to be used as a proxy for cerebral metabolism [
13]. However, there are some limitations to assessment with fMRI; namely it does not provide a direct measure of neuronal oxygen consumption or neuronal activation and results are vulnerable to movement distortion. Conversely, radiolabelled water (H
2
15O) positron emission tomography (PET) is a direct measure of cerebral oxygen consumption (and thus neuronal activation) that is less affected by movement and more quantifiable than fMRI [
14]. This hypothesis-driven study aimed to test if cognitive performance (assessed by H
2
15O PET) and associated CBF estimates are affected at less pronounced levels of hypoglycaemia than previously studied and if cognitive performance is affected in the recovery phase following less pronounced hypoglycaemia.