Purpose-Gold is valued as an important commodity for investment. It provides stability in economic uncertainties and plays an important role in international trade and investment. This study attempts to find a model that fits well in predicting the prices of this yellow metal. Forecasting gold prices effectively is essential for investors, policymakers, and economic analysts alike, as it provides critical insights for decision-making in uncertain economic environments. This study aims to identify a robust ARIMA model that can accurately predict future gold prices. Design - The data related to gold prices is collected from www.worldgoldcouncil from the year 1978 till 2023. This 45-year data is tested for time series modeler ARIMA. To ascertain whether the data is stationary, an Augmented-Dickey fuller test is used. Visually evaluating the data’s stationarity was made easier by the Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF). ACF and PACF were then analyzed to visually assess the patterns in data, aiding in the preliminary selection of model parameters for further testing. Findings- 12 different models of ARIMA were tested, and ARIMA (0,1,1) was found the best fit based on the Bayesian Information Criterion (BIC). Among the different ARIMA models, the one with the lowest BIC is preferred. This model demonstrated superior forecasting accuracy by capturing the fundamental pattern and periodic fluctuations in gold prices effectively. Originality- The study is based on the application of long-term historical data, using a 45-year dataset from 1978 to 2023. This data provides depth and richness to the analysis, capturing various economic cycles, crises and trends that affect gold prices.