BACKGROUND: Discrepancy between mammographic and pathological sizing of DCIS can lead to surgical overtreatment, with poorer cosmesis or unnecessary mastectomy, or undertreatment and recurrence. METHODS: Within the UK Sloane Project prospective cohort study of screen-detected DCIS (2003-2012), we investigated factors associated with 'pathology larger (PL)' (pathological larger than mammographic size) or 'mammogram larger (ML)' (mammographic larger than pathologic size), size discrepancy and the impact on ipsilateral recurrence. RESULTS: Among 9937 patients (mean age 60; range 46-87), mammographic size remained constant at median 19 mm (IQR 10-35)mm whilst pathological size increased from 16(10-28)mm to 20(10-33)mm (p = 0.001)over the study. The mammographic and pathological size discrepancy decreased from 3.4 mm to 0.2 mm (p < 0.05). In patients undergoing BCS, size discrepancy of ≥5 mm was associated with increased 5-year ipsilateral recurrence if lesions were PL (odds ratio(OR) 1.37 (C.I. 1.03-1.82, p = 0.03) and if lesions were ML (OR 1.4 (C.I. 1.10-1.86, p = 0.008), compared to <5 mm discrepancy. Factors associated with PL by ≥ 5 mm were high grade (OR 1.9 [95 % CI 1.5-2.4, p < 0.001]) and mastectomy (OR 4.4 [C.I. 3.8-5.1, p < 0.001]) and for ML ≥ 5 mm was larger mammographic tumour size (>40 mm; OR 115.7 [C.I. 82.3-162.6], p < 0.001]). CONCLUSION: Mammographic-pathological size discrepancy is associated with higher recurrence following BCS for DCIS.