We report an evaluation of current laboratory practice for the diagnosis of von Willebrand's disease (VWD) by means of a multilaboratory survey. This assessment was undertaken with the RCPA Quality Assurance Program (QAP) in Haematology, which covers a wide geographic area encompassing Australia, New Zealand and Asia. A total of 25 laboratories actively involved in testing for VWD were selected to participate in a sample testing assessment exercise. Samples comprised 10 plasmas: (i) a normal plasma pool (in duplicate), (ii) this pool diluted to 50% (in duplicate), (iii) a normal individual (X1), (iv) severe Type 1 VWD (X1), (v) Type 2B VWD (x2 unrelated donors), (vi) Type 3 VWD (x1), (vii) Type 2A VWD (x1). Laboratories were asked to perform all tests available to them in order to establish a laboratory diagnosis of VWD, and then to comment on the possibility or otherwise of VWD. Overall findings indicated a wide variation in test practice, in the effectiveness of various test procedures in detecting VWD, and in the ability of various composite test panels to identify type 2 VWD subtypes. Firstly, while all laboratories (n = 25) performed tests for FVIII:C activity, von Willebrand factor 'antigen' (VWF:Ag) and a functional VWF assay [using the ristocetin cofactor assay (VWF:RCo; n = 23) and/or the collagen binding assay (VWF:CBA; n = 12)], only three laboratories carried out VWF:Multimer analysis. Secondly, for the three quantitative VWF assays, 10/25 (40%) laboratories performed all three, whereas 15/25 (60%) performed only two [VWF:Ag and VWF:RCo (n = 13); VWF:Ag and VWF:CBA (n = 2)]. Thirdly, a variety of assay methodologies were evident for VWF:Ag [ELISA, electro-immuno diffusion (EID), latex immuno-assay (LIA), and VIDAS assay] and VWF:RCo (platelet agglutination/'aggregometry' and a 'functional VWF:RCo-alternative' ELISA assay). Between method analysis for the quantitative VWF assays showed that the VWF:RCo yielded the greatest degree of inter-laboratory assay variation, and had the poorest overall performance with respect to sensitivity to low levels of VWF. The VWF:CBA also performed better than the VWF:RCo in terms of ability to detect functional VWF 'discordance' (i.e. Type 2 VWD). Within VWF:Ag method analysis showed that the EID assay procedure was associated with the greatest variation in assay results, while the EID and LIA test methods showed poorer sensitivity at low VWF levels compared to the ELISA method. Within the VWF:RCo assay procedure, greatest variation in assay results and poorest sensitivity to low VWF levels was obtained using the agglutination method; however, the agglutination procedure showed better performance than the 'functional VWF:RCo-alternative' ELISA assay in identifying Type 2 VWD plasma samples. Finally, despite identified variations, most laboratories appeared to understand the complexities involved in the VWD-diagnostic process, and made appropriate diagnostic predictions regarding tested samples. From a total possible 246 interpretation events, laboratories in most cases correctly identified normal samples as normal (67/75 events = 89%), and VWD samples as derived from individuals with VWD (117/121 events = 97%). Moreover, when VWD was suggested by laboratory findings, laboratories usually correctly predicted the general subtype of VWD present (96/109 events = 88%). When 'misinterpretations' occurred, these could often be linked to the test panels utilised by laboratories. That is, laboratories using the VWF:Ag and VWF:RCo combination were more likely to incorrectly identify samples derived from Type 2 VWD patients as being Type 1, Type 1 VWD patients as being Type 2, and normal plasma samples as potentially derived from patients with VWD, compared to those using the VWF:Ag and VWF:CBA.