This study was performed to estimate total contents of arsenic (As) by stepwise multiple-regression analysis using chemical properties and extractable contents of metal in paddy soil adjacent to abandoned mines. The soil was collected from paddies near abandoned mines. Soil pH, electrical conductively (EC), organic mater (OM), available phosphorus ($P_2O_5$), and exchangeable cations (Ca, K, Mg, Na) were measured. Total contents of As and extractable contents of metals were analyzed by ICP-OES. From stepwise analysis, it was showed that the contents of extractable As, available phosphorus, extractable Cu, exchangeable K, exchangeable Na, and organic mater significantly influenced the total contents of As in soil (p<0.001). The multiple linear regression models have been established as Log (Total-As) = 0.741 + 0.716 Log (extractable-As) - 0.734 Log (avail-$P_2O_5$) + 0.334 Log (extractable-Cu) + 0.186 Log (exchangeable-K) - 0.593 Log (exchangeable-Na) + 0.558 Log (OM). The estimated value in total contents of As was significantly correlated with the measured value in soil ($R^2$=0.84196, p<0.0001). This predictive model for estimating total As contents in paddy soil will be properly applied to the numerous datasets which were surveyed with extractable heavy metal contents based on Soil Environmental Conservation Act before 2010.