In this study, new prediction equations for significant duration ( and ) are developed using an Iranian strong ground‐motion database. The database includes 2228 records of 749 earthquakes with small to large magnitudes up to the year 2018. The functional form of the model is an additive natural logarithm of four predictor variables, namely moment magnitude (), rupture distance (), time‐averaged shear‐wave velocity in the top 30 m (), and the style of faulting effect (), which is considered as an indicator directly in the functional form for the first time. The proposed models can be used to estimate significant durations of earthquakes with moment magnitudes () from 4.5 to 7.6 and rupture distances of up to 200 km. The models are compared with four existing significant‐duration prediction models. The results indicate proper agreement between the proposed models and the models that use the Pacific Earthquake Engineering Research Center‐Next Generation Attenuation‐West2 Project (PEER‐NGA West2) database (say PEER models). Based on the results, our proposed models indicate an increasing trend of significant duration with an increase in the rupture distance. However, unlike the PEER models, the rate of increase in significant duration is decreasing in our model.