Over 1500 soils samples have been analyzed for Cd. Samples were 514 soils taken in such a way as to cover a wide range of soil types common to Japan (referred to as nationwide samples), 139 volcanic ash soils also taken nationwide scale (volcanic ash samples), and 887 soils taken from arable lands in Miyagi Prefecture, northeastern Japan (Miyagi samples). Histogram has revealed that the frequency distributions of Cd was positively skewed and coincided well with those of log normal distributions, indicating arithmetic mean value is not appropriate to represent the Cd status in soils. The anti-log values of the minimum, mean, maximum, and 95% confidence limit of the mean calculated using log transformed data were respectively 0.015, 0.27, 3.37 and 0.06〜1.09mg kg^<-1>. Whereas the higher outliers in Miyagi samples were polluted soils, those in nationwide samples were un-polluted dark red soils (Chromic Luvisols) and red soils (Orthic Acrisols) both derived from limestone. It is assumed that trace amounts of Cd contained in the parent materials as impurities at the initial stage of weathering were gradually concentrated during the succeeding weathering processes as almost all of CaCO_3 were lost. The above hypothesis is strongly exemplified in the findings that the concentration levels of more than 30 trace elements in these soils were also higher than those of the other soils. It is worth mentioning that the occurrence of soil samples containing more than 3mg kg^<-1> of Cd not necessarily indicates events related to the anthropogenic soil pollution. The concentration range of Cd in volcanic ash samples was apparently lower than that of the other two groups. Comparison of concentration levels of Cd between volcanic ash soils and non-volcanic ash soils after excluding outliers has revealed that Cd in the former were significantly lower than that in the latter.