To identify the efficacy of early cerebral MR, performed in the first month of birth, in the detection of brain lesions in high risk preterm infants, compared with conventional US, we recruited into the study a group of 30 preterm infants born at or below a gestational age of 30 weeks, who had a pathologic scan. The findings on US were compared with those of the early MR scan, performed in the same days, the results of which were considered as the final diagnosis. The value of cranial US as a predictor of MR signal intensity was assessed by calculating sensitivity, specificity, positive and negative predictive values. Agreement between two investigations was evaluated by calculating the K coefficient. US showed 33 haemorrhagic lesions in 25 preterms; MR showed 27 haemorrhagic lesions in 22 infants: in 16 cases MR gave the same results of US. Cranial US was reliable in detecting lesions such as GLH and IVH, but less sensitive in the definition of their size and distribution. Sensitivity of US for haemorrhagic lesions was 96.3%, PPV 78.8%, K coefficient 0.55 (p < 0.001). About the White Matter, cranial US demonstrated 20 lesions in 20 preterms; MR showed 16 lesions in 16 infants: in 3 cases MR was agree to US. US showed high reliability in the detection of cystic lesions, but significant limitations in the demonstration of non-cystic injury. We founded that normal WM echogenicity on US is not a good predictor of normal WM signal intensity on MR (30%). Sensitivity of US for WM lesions was 81.3%, PPV 65%, K coefficient 0.23 (p = 0.04). Finally US showed 4 lesions in other brain locations, MR confirmed 3 of them and discovered other 10. Sensitivity of US for these lesions was 23.1%, PPV 75%, K coefficient 0.21 (p = 0.11). We founded that cranial US is a good method for detecting GLH, IVH, HPI and severe WM lesions (cystic PVL), but it can miss non-cystic PVL, punctate haemorrhages, WMD and lesions in other brain locations, that, on the other hand, MR detects clearly.