In this paper, we propose a simple model-based time-of-arrival (TOA) estimation technique for multi-band orthogonal frequency division multiplexing (MB-OFDM) signals based on the ECMA-368 standard . The proposed technique does not need oversampling at the receiver. The key idea is to minimize the energy leakage from the first channel path due to mis-sampling. We use realistic channel models and assume that no channel statistical information is known at the receiver. In our approach, the multi-band signals are coherently combined in order to improve the resolution of the TOA estimation. We compare the performance of the proposed scheme with the well-known maximum-likelihood based algorithm, namely, space-alternating generalized expectation-maximization (SAGE). Simulations based on the ECMA-368 standard parameters in both IEEE 802.15.3a and IEEE 802.15.4a standard channel models show that our approach outperforms the SAGE algorithm by avoiding the local maxima. Furthermore, we have also shown that our approach can achieve a root mean square error of TOA estimation much smaller than the receiver sampling interval and it is robust to the narrowband interference.