Background  There is increasing reliance on consensus criteria for decision making. Recent criteria state that acromegaly is excluded by a nadir GH during an oral glucose tolerance test (OGTT) of < 1 µg/l and a normal level of IGF‐I.

Objective  To study GH and IGF‐I assay performance close to cut‐off values for active acromegaly.

Design and methods  Two serum samples known to give borderline results were sent to all centres participating in the UK National External Quality Assessment Service (NEQAS). Sample A was assigned to be a nadir during an OGTT and sent for GH assessment to 104 centres. Sample B, with a clinical scenario, was sent to 23 centres that measure IGF‐I, and these centres were asked to measure IGF‐I, interpret the result and provide the source of their reference ranges (RRs).

Results  For sample A, the median GH was 2·6 mU/l (range 1·04–3·5 mU/l). Applying a conversion factor (CF) of 2·0 (1 µg/l = 2 mU/l), the most negatively biased method classified 10% of the values consistent with acromegaly, while the most positively biased method classified all values as consistent with the diagnosis. Applying a CF of 3·0 (1 µg/l = 3 mU/l), only 11% of results were consistent with acromegaly. For sample B, the median IGF‐I was 50·8 nmol/l (range 24·3–60·9 nmol/l). All centres used age‐related RRs. There was a 50% variation in the upper limit of the RRs between centres. Overall, 30% of the IGF‐I results were against the diagnosis. There was little agreement in the RRs quoted by centres using the same method.

Conclusion  Variability in assay performance, coupled with use of inappropriate CFs and RRs, undermines the applicability of international consensus criteria to local practice.