A robust approach for estimating the expected maximum levels of radiofrequency electromagnetic emissions is proposed. The expected maximum peak value is intended to provide a statistical approximation for the worst case emissions scenario that accounts for the variability of the measured interference. The estimates are obtained through Monte Carlo resampling from a non-parametric distribution fitted by means of kernel density estimation applied to the time-frequency representation of the assessed interference. As a key advantage, calculating the expected maximum peak value does not require increasing the dwell time or holding the maximum value over successive sweeps. Results indicate the methodology is better suited than previous approaches for calculating the expected maximum peak value because it does not depend on normality assumptions difficult to guarantee in practice. The proposed technique is an example of how full time-domain EMI measurements can be exploited for obtaining further insights.