Background: Cerebral amyloid angiopathy (CAA) is an age-related small vessel disease, characterised pathologically by progressive deposition of amyloid β in the cerebrovascular wall. The Boston criteria are used worldwide for the in-vivo diagnosis of CAA but have not been updated since 2010, before the emergence of additional MRI markers. We report an international collaborative study aiming to update and externally validate the Boston diagnostic criteria across the full spectrum of clinical CAA presentations. Methods: In this multicentre, hospital-based, retrospective, MRI and neuropathology diagnostic accuracy study, we did a retrospective analysis of clinical, radiological, and histopathological data available to sites participating in the International CAA Association to formulate updated Boston criteria and establish their diagnostic accuracy across different populations and clinical presentations. Ten North American and European academic medical centres identified patients aged 50 years and older with potential CAA-related clinical presentations (ie, spontaneous intracerebral haemorrhage, cognitive impairment, or transient focal neurological episodes), available brain MRI, and histopathological assessment for CAA diagnosis. MRI scans were centrally rated at Massachusetts General Hospital (Boston, MA, USA) for haemorrhagic and non-haemorrhagic CAA markers, and brain tissue samples were rated by neuropathologists at the contributing sites. We derived the Boston criteria version 2.0 (v2.0) by selecting MRI features to optimise diagnostic specificity and sensitivity in a prespecified derivation cohort (Boston cases 1994–2012, n=159), then externally validated the criteria in a prespecified temporal validation cohort (Boston cases 2012–18, n=59) and a geographical validation cohort (non-Boston cases 2004–18; n=123), comparing accuracy of the new criteria to the currently used modified Boston criteria with histopathological assessment of CAA as the diagnostic standard. We also assessed performance of the v2.0 criteria in patients across all cohorts who had the diagnostic gold standard of brain autopsy. Findings: The study protocol was finalised on Jan 15, 2017, patient identification was completed on Dec 31, 2018, and imaging analyses were completed on Sept 30, 2019. Of 401 potentially eligible patients presenting to Massachusetts General Hospital, 218 were eligible to be included in the analysis; of 160 patient datasets from other centres, 123 were included. Using the derivation cohort, we derived provisional criteria for probable CAA requiring the presence of at least two strictly lobar haemorrhagic lesions (ie, intracerebral haemorrhages, cerebral microbleeds, or foci of cortical superficial siderosis) or at least one strictly lobar haemorrhagic lesion and at least one white matter characteristic (ie, severe visible perivascular spaces in centrum semiovale or white matter hyperintensities in a multispot pattern). The sensitivity and specificity of these criteria were 74·8% (95% CI 65·4–82·7) and 84·6% (71·9–93·1) in the derivation cohort, 92·5% (79·6–98·4) and 89·5% (66·9–98·7) in the temporal validation cohort, 80·2% (70·8–87·6) and 81·5% (61·9–93·7) in the geographical validation cohort, and 74·5% (65·4–82·4) and 95·0% (83·1–99·4) in all patients who had autopsy as the diagnostic standard. The area under the receiver operating characteristic curve (AUC) was 0·797 (0·732–0·861) in the derivation cohort, 0·910 (0·828–0·992) in the temporal validation cohort, 0·808 (0·724–0·893) in the geographical validation cohort, and 0·848 (0·794–0·901) in patients who had autopsy as the diagnostic standard. The v2.0 Boston criteria for probable CAA had superior accuracy to the current Boston criteria (sensitivity 64·5% [54·9–73·4]; specificity 95·0% [83·1–99·4]; AUC 0·798 [0·741–0854]; p=0·0005 for comparison of AUC) across all individuals who had autopsy as the diagnostic standard. Interpretation: The Boston criteria v2.0 incorporate emerging MRI markers of CAA to enhance sensitivity without compromising their specificity in our cohorts of patients aged 50 years and older presenting with spontaneous intracerebral haemorrhage, cognitive impairment, or transient focal neurological episodes. Future studies will be needed to determine generalisability of the v.2.0 criteria across the full range of patients and clinical presentations. Funding: US National Institutes of Health (R01 AG26484).