As conventional road traffic noise prediction models are designed to estimate long-term representative noise levels, capturing fine-scale noise fluctuations caused by real-world traffic dynamics is challenging. A previous study proposed a microscopic road traffic noise model (MTN) can calculate time-series noise levels with a resolution of 1 s using the concept of a moving noise source. In this study, two experiments were conducted to verify the accuracy of the noise prediction of the model. First, by comparing the calculated noise levels of two conventional road traffic noise models and the MTN in a simple road simulation environment, it was confirmed that the calculation error was within 3 dB(A) when calculating the 1-h equivalent noise level. Second, an experiment was conducted to verify the noise prediction error of the MTN on six actual roads. A comparison of the calculated noise level using the MTN based on traffic data collected from actual roads with the measured noise level on real roads showed that the calculated noise level achieved a mean absolute error (MAE) of 1.88 dB(A) from the equivalent noise level and 1.28 dB(A) from the maximum noise level. This was similar to the MAE of the foreign road traffic noise models. However, when the location of the receiver is within 10 m of the road, an error of more than 3 dB(A) occurs because of the simplicity of the MTN propagation model, which remains a problem that must be solved in the future. This study proved that the noise level calculation using the MTN is similar to the noise of an actual road environment. Additionally, the continuous development of the MTN is expected to make it an effective alternative for the management of road noise.