Magnitude-charge size relations have important uses in forensic seismology and are used in Comprehensive Nuclear-Test-Ban Treaty monitoring. I derive empirical magnitude versus detonation-charge-size relationships for 322 detonations located by permanent seismic networks in California and Nevada. These detonations, used in 41 different seismic refraction or network calibration experiments, ranged in yield (charge size) between 25 and 106 kg; coda magnitudes reported for them ranged from 0.5 to 3.9. Almost all represent simultaneous (single-fired) detonations of one or more boreholes. Repeated detonations at the same shotpoint suggest that the reported coda magnitudes are repeatable, on average, to within 0.1 magnitude unit. An empirical linear regression for these 322 detonations yields M = 0.31 + 0.50 log10(weight [kg]). The detonations compiled here demonstrate that the Khalturin et al. (1998) relationship, developed mainly for data from large chemical explosions but which fits data from nuclear blasts, can be used to estimate the minimum charge size for coda magnitudes between 0.5 and 3.9. Drilling, loading, and shooting logs indicate that the explosive specification, loading method, and effectiveness of tamp are the primary factors determining the efficiency of a detonation. These records indicate that locating a detonation within the water table is neither a necessary nor sufficient condition for an efficient shot.