Determining the false negative rate of cervical smear interpretation is an important part of quality assessment and a necessary step for any improvement program. We report our experience of negative smear rescreening of 522 histologically proven high-grade lesions or cancers, over a 5 to 7 year preceding period. False negative rate was 6.88% as calculated with a narrow definition of error, i.e. intra-epithelial lesions and atypical squamous cells of undetermined significance. It was 10.78% as calculated with a broad definition of error, including minor anomalies such as repair and parakeratosis. Bibliographic data account for 0 to 94% false negative diagnoses, owing to great disparities in calculating the false negative rate as well as in rescreening. However, a 10% traditionally calculated and standardised false negative rate is a reasonable and achievable goal in a view of quality improvement. Systematic random rescreening of 10% of negative smears is ineffective.