In several recent epidemiological studies blood lead levels (BLLs) even below the current CDC intervention level of 10 microg/dl have been associated with reduced neurocognitive capacities of children, with no clear evidence of a "safe" threshold. We analyzed the relationship between the BLLs and the neurocognitive capacities of 205 Sardinian students aged 11 to 15 years, using 2 tests of the Swedish Performance Evaluating System (SPES) and the full-scale Intelligence Quotient (IQ) derived from the Wechsler Intelligence Scale for Children (WISC). The studied population included 104 children (61 males and 43 females) living in Portoscuso, a town 2 Km far from a lead smelter (mean BLLs: 5.98 +/- 2.2; max 11.5 microg/dl), and 101 age-matched students (55 males and 46 females) living in Sant'Antioco, a town about 20 Km far from the same smelter (mean BLLs: 2.08 +/- 0.8; max 4.5 microg/dl). Subjects with BLLs above 4 microg/dl performed worse in the SPES tests and scored about 5.0 points less on the full-scale IQ compared to the students with lower BLLs. The adjusted regression coefficients derived from the multivariate analysis showed that higher BLLs were significantly associated with worse performances in the SPES tests and with reduced IQ (0.94 points for each microg/dl of BLL). This study confirms the potential neurotoxicity of low-levels of lead suggesting the need of lowering the actual CDC "limit of concern" for children to values lower than 4 microg/dl, improving at the same time the environmental primary prevention for limiting the lead exposure of subjects living near the lead smelter.