© yyyy IAHS Press Hydrological Sciences Journal – Journal des Sciences Hydrologiques 00 (00) 2016 10.1080/02626667.yyyy.nnnnnn Comparison on performance of artificial intelligence, statistical and time series methods for flood forecasting Nuruljannah Khairuddin 1, Ahmad Zaharin Aris 1,* , Ahmed El-Shafie 2 , Tahoora Sheikhy Narany 1 1 Faculty of Environmental Studies, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia 2 Department of Civil Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia * Corresponding author: Phone: +603-89467455; Fax: +603-89467463; Email: zaharin@upm.edu.my Received Editor Abstract Flood forecasting is an important tool for river management as early flood warning system. The aims of this study are to develop statistical, time series and artificial intelligence methods, to investigate the implementation of each time-series technique in order to find effective tool for water level prediction in flood forecasting. This paper explores the application of water level and rainfall data and input time series into three different methods; Linear Regression, Auto-Regressive Integrated Moving Average (ARIMA) and Artificial Neural Networks (ANN) models in Muda River in Malaysia. The performances of the models were compared based on the maximum correlation coefficient and minimum Root Means Square Error. Based on the results, ANNs model presents the accurate measured with R value of 0.868 and 15% of percentage error. The present study suggests that ANNs is the best model to its ability in recognizing times series pattern and understand well in non-linear relationships. Key words Linear Regression; ARIMA; Artificial Neural Networks; Flood Forecasting