Introduction Neurotoxicity due to acute prenatal exposure to high-dose of mercury (Hg) is well documented. However, the effect of prenatal exposure to low Hg levels on child neurodevelopment and the question about “safety” of fish-eating during pregnancy remain controversial. International comparisons of Hg concentrations in mother-child biological samples and neurodevelopmental scores embedded in birth cohort studies may provide useful evidence to explore this issue. Materials and methods The Mediterranean (Italy, Slovenia, Croatia, and Greece) cohort study included 1308 mother-child pairs enrolled in the Public Health Impact of long-term, low-level, Mixed Element exposure in a susceptible population EU Sixth Framework Programme (PHIME). Maternal hair and venous blood, cord blood and breast milk samples were collected, and total Hg (THg) levels were measured. Demographic and socioeconomic information, lifestyles and nutritional habits were collected through questionnaires at different phases of follow-up. Children at 18 months of age underwent neurodevelopmental testing using the Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III). Multivariate linear and logistic regressions were performed, for each country, to assess the association between THg and BSID-III scores, obtaining adjusted β coefficients and odds ratios (ORs). These values were used to conduct a meta- analysis, to explore possible heterogeneity among countries and to obtain combined estimates of the association between THg exposure and BSID-III scores. Results Median THg (ng/g) was: 704 in maternal hair, 2.4 in maternal blood, 3.6 in cord blood, and 0.6 in breast milk. THg concentrations were highest in Greece and lowest in Slovenia. BSID-III neurodevelopmental scores were higher in Croatia and Slovenia. The meta-analysis of multivariate linear models found an overall positive association between language composite score and receptive communication scaled score and increasing THg in maternal hair (n = 1086 ; β = 0.55 ; 95%CI: 0.05–1.05 and n = 1075 ; β = 0.12 ; 95%CI: 0.02–0.22, respectively). The meta-analysis of logistic regression models showed that the overall adjusted OR between THg in cord blood and suboptimal gross motor score was borderline significant (n = 882 ; OR = 1.03 ; 95%CI: 1.00–1.07). Heterogeneity was found across the four sub-cohorts for language composite score in maternal blood, and for fine motor scaled score in cord blood and breast milk. Language composite score and THg concentrations in maternal venous blood were positively related (n = 58 ; β = 4.29 ; CI95% (−0.02, 8.60)) in Croatia and an increase of 1 ng/g of THg in maternal venous blood was associated with a reduced risk for children to fall in the lowest quintile of language score by 31% (n = 58 ; OR = 0.69 ; CI 95%: 0.37, 1.01). The comparison of β coefficients obtained by multiple linear regression model showed an inverse association between fine motor score and THg concentrations in cord blood for Croatia (n = 54 ; β = −0.53 ; CI 95%: −1.10, 0.04) and Slovenia (n = 225 ; β = −0.25 ; CI 95%: −0.49, −0.01). In Slovenia THg level in breast milk was associated with suboptimal fine motor performance (n = 195 ; OR = 5.25 ; CI 95%: 1.36, 21.10). Conclusions This study showed an inverse relation between THg levels and developmental motor scores at 18 months, although the evidence was weak and partially internally and externally inconsistent. No evidence of detrimental effects of THg was found for cognitive and language outcomes at these concentrations and age.