The ML scale, introduced by Richter in 1935, is the antecedent of every magnitude scale in use today. The scale is defined such that a magnitude-3 earthquake recorded on a Wood-Anderson torsion seismometer at a distance of 100 km would write a record with a peak excursion of 1 mm. To be useful, some means are needed to correct recordings to the standard distance of 100 km. Richter provides a table of correction values, which he terms -log Ao, the latest of which is contained in his 1958 textbook. A new analysis of over 9000 readings from almost 1000 earthquakes in the southern California region was recently completed to redetermine the -log Ao values. Although some systematic differences were found between this analysis and Richter's values (such that using Richter's values would lead to underand overestimates of ML at distances less than 40 km and greater than 200 km, respectively), the accuracy of his values is remarkable in view of the small number of data used in their determination. Richter's corrections for the distance attenuation of the peak amplitudes on Wood-Anderson seismographs apply only to the southern California region, of course, and should not be used in other areas without first checking to make sure that they are applicable. Often in the past this has not been done, but recently a number of papers have been published determining the corrections for other areas. If there are significant differences in the attenuation within 100 km between regions, then the definition of the magnitude at 100 km could lead to difficulty in comparing the sizes of earthquakes in various parts of the world. To alleviate this, it is proposed that the scale be defined such that a magnitude 3 corresponds to 10 mm of motion at 17 km. This is consistent both with Richter's definition of ML at 100 km and with the newly determined distance corrections in the southern California region. Aside from the obvious (and original) use as a means of cataloguing earthquakes according to size, ML has been used in predictions of ground shaking as a function of distance and magnitude; it has also been used in estimating energy and seismic moment. There is a good correlation of peak ground velocity and the peak motion on a Wood-Anderson instrument at the same location, as well as an observationally defined (and theoretically predicted) nonlinear relation between ML and seismic moment. An important byproduct of the establishment of the ML scale is the continuous operation of the network of Wood-Anderson seismographs on which the scale is based. The records from these instruments can be used to make relative comparisons of amplitudes and waveforms of recent and historic earthquakes; furthermore, because of the moderate gain, the instruments can write onscale records from great earthquakes at teleseismic distances and thus can provide important information about the energy radiated from such earthquakes at frequencies where many instruments have saturated. ?? 1989.