<p>Reliable magnitude estimation is essential for earthquake early warning systems. This study develops empirical magnitude estimation relations for Iran, based on frequency, amplitude, and energy characteristics of the P-wave onset using earthquakes waveforms recorded across the country. The P-wave parameters of characteristic period (τ_c), dominant period (τ_p^max), peak displacement (P_d), velocity, and acceleration, cumulative absolute velocity, integral of velocity squared, and integral of absolute velocity are calculated within initial P-wave windows of 1–8&#xa0;s. Linear regression models are developed and tuned based on 80% of the dataset and then validated on the remaining 20%. The analysis of the P-wave time window effect shows that the conventional 3-second window is the shortest duration at which errors noticeably decrease and model performance improves, while for τ_c, the scatter reduces more from the 5-second window. The results indicate that magnitude estimation relations based on amplitude- and energy-related parameters generally provide more stable and accurate performance, with lower standard deviation and scatter, compared to those based on frequency parameters. Among all parameters, relations derived from P_d exhibit the smallest errors and scatter, whereas those based on τ_p^max (within the 3-second window) show the least magnitude underestimation for earthquakes larger than M 6.5. Finally, comparisons with global and regional models highlight both compatibilities and also significant differences, particularly for energy-based parameters, underlining the importance of regional calibration.</p>