The International Seismological Centre, in collaboration with the Global Earthquake Model effort, has released a new global earthquake catalog, covering the time period from 1900 through the end of 2009. In order to use this catalog for global earthquake studies, I determined the magnitude of completeness (M_c) as a function of time by dividing the earthquakes shallower than 60 km into 7 time periods based on major changes in catalog processing and data availability and applying 4 objective methods to determine M_c, with uncertainties determined by non-parametric bootstrapping. Deeper events were divided into 2 time periods. Due to differences between the 4 methods, the final M_c was determined subjectively by examining the features that each method focused on in both the cumulative and binned magnitude frequency distributions. The time periods and M_c values for shallow events are: 1900-1917, M_c=7.7; 1918-1939, M_c=7.0; 1940-1954, M_c=6.8; 1955-1963, M_c=6.5; 1964-1975, M_c=6.0; 1976-2003, M_c=5.8; and 2004-2009, M_c=5.7. Using these M_c values for the longest time periods they are valid for (e.g. 1918-2009, 1940-2009,…) the shallow data fits a Gutenberg-Richter distribution with b=1.05 and a=8.3, within 1 standard deviation, with no declustering. The exception is for time periods that include 1900-1917 in which there are only 33 events with M≥ M_c and for those few data b=2.15±0.46. That result calls for further investigations for this time period, ideally having a larger number of earthquakes. For deep events, the results are M_c=7.1 for 1900-1963, although the early data are problematic; and M_c=5.7 for 1964-2009. For that later time period, b=0.99 and a=7.3.