In current policy and research, the sentencing guidelines grid is the starting point for analysis. The actual impact of presumptive sentencing guidelines, however, must be tested with research that examines decisionmaking discretion and disparity at every stage of processing for similar cases, including the charging stage. Research shows that jurisdictions with narrow, presumptive sentencing ranges have relatively small amounts of unexplained sentencing variation for similar cases. This leads to the conclusion that the guidelines are achieving their intent of reducing discretion and consequently sentencing disparity; however, if unfettered decisionmaking discretion is located at a processing stage prior to sentencing, disparity in sentencing is hidden from such research. Prosecutors' charging and plea negotiation decisions will determine which sentencing guidelines apply in a given case, which produces disparity in sentencing. The potential for decisionmaking discretion and disparity is also inherent in the construction of the sentencing guidelines themselves. Sentencing guidelines generally consist of a two-dimensional grid that maps criminal history and current offense information onto recommended sentencing ranges. Such a system seems to base sentencing on legitimate offender characteristics and measures of offense severity; however, if members of different racial groups are distributed differently across these legitimate sentencing factors, there is potential for the sentencing guidelines to institutionalize racially disparate punishment. More than 20 years ago, Fisher and Kadane (1983) noted efforts to reduce sentencing variation by developing guidelines based on past sentencing practices. They correctly argued that if past practices were discriminatory and if recommended punishments were based on these past practices without identifying the role of race, then the resulting punishment system would institutionalize racially disparate punishment. 7 tables and 30 references