We hypothesized that in patients with negative fluid-attenuated inversion recovery (FLAIR) images T(2) weighted fast spin-echo (FSE) images and T(1) weighted spin-echo (SE) images before and after intravenous administration of gadolinium-based contrast medium display no pathology either. Thus, we assessed the negative predictive value of FLAIR images to rule out MR-detectable brain lesions. 1026 consecutive cranial MR examinations were reviewed. Routine MRI of the brain included T(1) weighted coronal imaging before and after administration of gadopentetate dimeglumine, axial T(2) weighted FSE and fast-FLAIR imaging. The FLAIR images were rated by two radiologists into categories of 0 (without pathologic changes) and 1 (with pathologic changes). Two other radiologists analysed the complete examination. In 284 MR examinations of the brain no abnormalities were found (28%). FLAIR-ratings were false-negative in four cases and false-positive in 30 cases. Sensitivity and specificity of the FLAIR sequence for MR-detectable brain lesions were 99.5% and 89.4%. The unselective application of gadolinium avoided one false-negative MR-reading and improved the sensitivity of the MR-examination from 99.5% to 99.6%. Positive and negative predictive values were 96.1% and 98.4%, respectively. The interobserver reliability was kappa=0.93 for the FLAIR-readers and 0.89 for the readers who rated the complete examination. In conclusion, negative FLAIR images provide a high negative predictive value for MR-detectable brain lesions. Thus, in patients with negative FLAIR images the unselective application of gadolinium seems to be unnecessary.