Seismic moment, a measurable and well‐understood quantity of seismic sources, is used to estimate the yield of explosions. Application of such a method in the past, as in the manner of ‐derived yields, has been complicated by the effect of variations in the explosion working point, depth, and secondary source effects (such as spalling and tectonic release) on the observed moment. We start using the full (six‐element) moment tensor solution, which can capture the relevant source physics and, at least in theory, better isolate the primary explosion source. The moment‐to‐yield ratio is then estimated using an explosion source model which, provided with emplacement conditions, can relate the two parameters. We discuss the major sources of uncertainty associated with the method, and calibrate it with chemical and nuclear explosions at the Nevada National Security Site. We then apply the method to published moment tensor solutions for the six declared North Korean nuclear explosions that occurred between 2006 and 2017. The results are mostly consistent with other yield estimates made using a variety of high‐frequency methods. This technique is a new approach to estimating explosive yield and simple to implement, as much of the complexity is captured by the source models.