The accuracy of pavement performance prediction is a critical part of pavement management and directly influences maintenance and rehabilitation strategies. Many models with various specifications have been proposed by researchers and used by agencies. This study presents nine variables affecting pavement condition and analyses the accuracy of the Group Method of Data Handling (GMDH) and Adaptive Neuro Fuzzy Inference System (ANFIS) models in predicting pavement performance in short and long terms of a pavement life cycle. The nine effective variables are extracted from groups of pavement age, traffic conditions, environmental changes and pavement structure. The International Roughness Index (IRI) is used as the pavement performance index. Results show that GMDH with 1st, 2nd, 3rd and 4th-order polynomials is not able to predict pavement performance for short and long terms. However, the specified ANFIS models can accurately predict pavement performance in the short term, but not in the long term.