Recent randomised controlled trials (RCTs) in primary Sjögren's syndrome used the European League Against Rheumatism (EULAR) Sjögren's Syndrome Disease Activity Index (ESSDAI) as their primary endpoint. Given the heterogeneous and complex nature of primary Sjögren's syndrome, it might be more appropriate to also assess other clinically relevant disease features. We aimed to develop a novel composite endpoint for assessing treatment efficacy in patients with primary Sjögren's syndrome: the Composite of Relevant Endpoints for Sjögren's Syndrome (CRESS). A multidisciplinary expert team selected clinically relevant items and candidate measurements for inclusion in the composite score. For each measurement, cutoff points for response to treatment were chosen based on expert opinion, previously published data on minimal clinically important improvements, and trial data, primarily the week-24 data of the single-centre ASAP-III trial of abatacept versus placebo. CRESS was validated using data from three independent RCTs: one trial of rituximab (TRACTISS), one of abatacept (multinational trial), and one of tocilizumab (ETAP). We calculated the number and percentage of patients who were responders in the separate CRESS items, and the percentage of responders based on the total CRESS at the primary endpoint visits (week 48 for TRACTISS, week 24 for the other two trials). Patients with fewer than three items available for evaluating CRESS response were imputed as non-responders. Based on expert opinion, five complementary items were selected to assess response: (1) systemic disease activity by Clinical ESSDAI (less than 5 points); (2) patient-reported symptoms by EULAR Sjögren's Syndrome Patient Reported Index, assessed by a decrease of at least 1 point or at least 15% from baseline; (3) tear gland item by Schirmer's test and ocular staining score, assessed by an increase of at least 5 mm or decrease of at least 2 points, respectively, in patients with abnormal Schirmer's test or ocular staining score findings at baseline, or, in patients with normal baseline values, assessed by no change to abnormal for both; (4) salivary gland item, assessed by unstimulated whole saliva secretion (increase of at least 25%) and salivary gland ultrasonography (decrease of at least 25%); and (5) serology, assessed by rheumatoid factor (decrease of at least 25%) and IgG (decrease of at least 10%). Total CRESS response is defined as response on at least three of five items. Post-hoc assessment of phase 3 trial data showed that CRESS response rates at the primary endpoint visits were 60% (24 of 40) for abatacept versus 18% (seven of 39) for placebo (p<0·0001) in ASAP-III, 49% (33 of 67) for rituximab versus 30% (20 of 66) for placebo (p=0·026) in the TRACTISS trial, 45% (41 of 92) for abatacept versus 32% (30 of 95) for placebo (p=0·067) in the multinational abatacept trial, and 18% (10 of 55) for tocilizumab versus 24% (13 of 55) for placebo (p=0·48) in the ETAP trial. The CRESS is a feasible, well-balanced, composite endpoint for use in trials of primary Sjögren's syndrome. As a next step, the CRESS will require validation in a prospective RCT. None. For the Dutch translation of the abstract see Supplementary Materials section.