We have developed three methods for the automatic picking of first breaks that can be used for marine, dynamite, or vibroseis shot records: a modified Coppens's method, an entropy-based method, and a variogram fractal-dimension method. The techniques are based on the fact that the transition between noise and noise plus signal can be automatically identified by detecting rapid changes in a certain attribute (energy ratio, entropy, or fractal dimension), which we calculate within moving windows along the seismic trace. The application of appropriate edge-preserving smoothing operators to enhance these transitions allowed us to develop an automated strategy that can be used to easily signal the precise location of the first-arrival onset. Furthermore, we propose a mispick-correcting technique to exploit the benefits of the data present in the entire shot record, which allows us to adjust the trace-by-trace picks and to discard picks associated with bad or dead traces. As a result, the consistency of the first-break picks is significantly improved. The methods are robust under noisy conditions, computationally efficient, and easy to apply. Results using dynamite and vibroseis field data show that accurate and consistent picks can be obtained in an automated manner even under the presence of correlated noise, bad traces, pulse changes, and indistinct first breaks.