Road roughness is a broad term that incorporates everything from potholes and cracks to the random deviations that exist in a profile. To build a roughness index, road irregularities need to be measured first. Existing methods of gauging the roughness are based either on visual inspections or using one of a limited number of instrumented vehicles that can take physical measurements of the road irregularities. This paper more specifically focuses on the estimation of a road profile (i.e., along the "wheel track"). This paper proposes a solution to the road profile estimation using a wavelet neural network (WNN) approach. The method incorporates a WNN which is trained using the data obtained from a 7-DOF vehicle dynamic model in the MATLAB Simulink software to approximate road profiles via the accelerations picked up from the vehicle. In this paper, a novel WNN, multi-input and multi-output feed forward wavelet neural network is constructed. In the hidden layer, wavelet basis functions are used as activate function instead of the sigmoid function of feed forward network. The training formulas based on BP algorithm are mathematically derived and a training algorithm is presented. The study investigates the estimation capability of wavelet neural networks through comparison between some estimated and real road profiles in the form of actual road roughness.