In general, the nature of gold that acts as a hedge against inflation and its stable price over the course of the financial crisis has made it a unique commodity. Priceforecasts are a must for gold producers, investors and central bank to know the current trends in gold prices. Forecasting the future value of a variableis often done with time series analysis method. This study was conducted to determine the best model for forecasting gold commodity prices as well as forecasting world gold commodity prices in 2018 using Box-Jenkins approach. The data used in this study wasobtained from Investing.Com from 2015 until 2017. Thisstudy shows that ARIMA (1,1,1) is the best model to predict gold commodity prices based on Mean Absolute Percentage Error (MAPE). MAPE value for ARIMA (1,1,1) is 0.02%, where this value proves that forecasting using ARIMA (1,1,1) is the best forecasting becauseMAPE value is less than 10%.