Cryptocurrency trading, particularly Bitcoin, faces significant price fluctuations, necessitating accurate price prediction to support decision-making. This study aims to apply the Autoregressive Integrated Moving Average (ARIMA) model for short-term Bitcoin price forecasting. The choice of ARIMA is based on its ability to capture price trends using historical data. The research adopts the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology, which comprises five main phases: Business Understanding, Data Understanding, Data Preparation, Modeling, and Evaluation. The dataset used consists of historical Bitcoin USD prices from investing.com, encompassing 1,462 records with seven attributes: Date, Price, Open, High, Low, Vol, and Change, covering the period from May 5, 2020, to May 5, 2024. The ARIMA model applied to this dataset is evaluated using the Root Mean Square Error (RMSE) metric to measure prediction error, along with a stationarity test using the Augmented Dickey-Fuller (ADF) test. Evaluation results indicate that the ARIMA (1, 2, 1) model achieves an RMSE of 0.0712, demonstrating strong predictive accuracy. The ADF test reveals that the original data is non-stationary, requiring two rounds of differencing (d=2) to achieve stationarity. Furthermore, analysis of the Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) confirms that the ARIMA (1, 2, 1) parameters are suitable for capturing the data patterns. Although the model performs well, the findings also suggest the need for more advanced predictive models, such as machine learning or deep learning techniques, to better handle higher market volatility.