Purpose: To determine the frequency, pathology and causes of a delay in cancer diagnosis in women recalled for suspicious screening mammography. Methods: We included all 290,943 screening mammograms of women aged 50-75 years, who underwent biennial screening mammography between 1st January 1995 and 1st January 2006. During a follow-up period of at least 2 years, clinical data, breast imaging reports, biopsy results and breast surgery reports were collected of all 3513 women with a positive screening result. Tumour stages of breast cancers with a diagnostic delay (defined as breast cancer confirmation more than 3 months following a positive mammography screen) were compared with those of cancers diagnosed within 3 months following referral and with interval cancers. Results: A diagnostic delay occurred in 97 (6.5%) of 1503 screen-detected cancers. These 97 false-negative assessments comprised significantly more ductal cancers in situ (26.8%) than did cancers with an adequate assessment after recall (15.5%, p = 0.004) or interval cancers (3.7%, p < 0.001). Compared with interval cancers, cancers with a false-negative assessment had a more favourable tumour size (T1a-c, 87.3% versus T1a-c, 46.4%; p < 0.001) and showed significantly fewer cases with axillary lymph node metastases (22.5% versus 48.2%; p < 0.001). Between hospitals having performed the workup of at least 500 referred women each, the percentage of women with a false-negative assessment varied from 5.0% to 9.1% (p = 0.03). In these hospitals, improper classification of lesions at diagnostic mammography comprised 64.4% of false-negative assessments. Conclusion: We found that 6.5% of recalled women experienced a delay in breast cancer diagnosis, with significant performance variations between hospitals. (C) 2008 Elsevier Ltd. All rights reserved.