This study aimed to determine the strength of association between anti-beta-2-glycoprotein I (anti-β2GPI) isotypes and thrombotic events by phenotyping hospital patients tested for diagnostic purposes with at least one positive anti-β2GPI isotype. A laboratory database search identified patients who had undergone anti-β2GPI testing during a 3 year period. Medical records of patients with a positive anti-β2GPI result were reviewed and clinical events ascertained. Thromboses were subdivided into provoked and unprovoked, depending on the stated aetiology. A total of 128 patients had at least one positive anti-β2GPI isotype. There was a higher proportion of unprovoked thromboses among patients who were IgG antiβ2GPI positive compared to those who lacked IgG anti-β2GPI (20/30 versus 20/98). Median IgG anti-β2GPI levels were higher among patients with unprovoked events compared to those without (22.5 SGU versus 2 SGU). Retrospective assessment of anti-β2GPI testing strategies showed that testing IgM and/or IgA anti-β2GPI after IgG anti-β2GPI captured a greater number of non-thrombotic events and provoked thromboses than unprovoked thromboses. IgG anti- β2GPI associates most strongly with clinical events characteristic of antiphospholipid syndrome (APS). These results suggest that IgG anti-β2GPI is superior to IgM and IgA anti- β2GPI in the assessment of hospital patients with potential APS. © 2014 Royal College of Pathologists of Australasia.