To analyze the relationship between gestational age and apparent diffusion coefficient (ADC) values in different regions of fetal brain from middle to late trimester. DW images performed in 70 singleton non-sedated fetuses with questionably abnormal results on sonography and normal fetal MR imaging results were retrospectively reviewed. The median gestational age was 32.4 weeks (range, 24-38).With the formula of ADC = ln (S600/S0)/(B0-B600), the mean ADC values were obtained for fetal parietal white matter (WM), frontal WM, temporal WM, occipital WM, pons, cerebellum, basal ganglia and thalamus. The relationship of mean ADC values in different regions with gestational age was analyzed with linear regression. The mean ADC values were 1.77 ± 0.32 mm(2)/s (SD) in fetal parietal white matter (WM), 1.71 ± 0.32 mm(2)/s in occipital WM, 1.31 ± 0.18 mm(2)/s in thalamus, 1.34 ± 0.15 mm(2)/s in basal ganglia. And the mean ADC values in cerebellum, pons, frontal WM and temporal WM were 1.17 ± 0.16, 1.41 ± 0.18, 1.87 ± 0.18 and 1.74 ± 0.19 mm(2)/s respectively. A significant negative correlation between ADC values and gestational age was found for parietal WM, occipital WM, pons, cerebellum, basal ganglia and thalamus (P < 0.05). Frontal WM ADC (P > 0.05) and temporal WM ADC (P = 0.05) did not significantly change with gestational age whereas only a downward trend was present. The correlation coefficient (r) and coefficient of regression (b) were 0.420 and -0.045 in parietal WM; 0.470 and -0.052 in occipital WM; 0.370 and -0.027 in cerebellum; 0.027 and -0.020 in pons; 0.320 and -0.027 in thalamus; 0.300 and -0.021 in basal ganglia. The mean ADC values peaked in frontal WM and lowest in pons. The mean ADC values in white matter were higher than those in deep gray nuclei, cerebellum and pons.With the development of fetal brain,ADC values declined the fastest in cerebellum and occipital WM, followed by basal ganglia and thalamus. Regional differences in non-sedated fetal brain ADC values and their evolutions with gestational age are likely to reflect variations in brain maturation.