Results indicated that the National Incident-Based Reporting System (NIBRS), with details on the offenses, victims, arrestees, and offenders, can address research that has relied on the use of offense, victim, or arrestee rates to approximate offender rates. The findings demonstrate that for the race-specific analysis of violent offenses, both victim and offense-known rates may be poor substitutes for an offender rate. The NIBRS offender rate is the most logical and useful offender measure because it provides a direct indication of the age, race, and sex of most offenders whether or not an arrest occurs. However, when it is not available, as it is not in the traditional Uniform Crime Report (UCR) program, an arrest measure provides a reasonable but not perfect substitute. Moreover, analysis of NIBRS arrest data for 2002 to 2004 indicates that modified arrest rates provide a closer approximation to pure offender rates than unmodified arrest rates. This further suggests that when offender rates are not available, the use of modified arrest rates is a more accurate measure of offending than either victimization rated or unmodified arrest rates. For race-specific measures even the best murder or robbery victim rate is a poor substitute for an offender rate, and offense-known measures are even poorer measures of offending. In addition, and unexpectedly, it was found that race-specific victim and offender counts and rates take much of the mystery out of the frequently used and variously interpreted variable percent Black. Data were collected using NIBRS 2002 to 2004 robbery and murder counts and census 2000 data for 166 cities. Tables, references