BJA collects an array of data on adult drug-court grantees, such as drug-court completion rates, and during the course of GAO's review, began expanding its use of this performance data to inform grant-related decisions, such as allocating resources and setting program priorities. For example, during September 2011, BJA assessed a sample of adult drug-court grantees' performance across a range of variables, using a new process it calls GrantStat. BJA developed recommendations following this assessment and is determining their feasibility. In addition, in October 2011, BJA finalized revisions to the performance measures on which grantees report. BJA's process of revising its performance measures generally adhered to key practices, such as obtaining stakeholder involvement; however, BJA could improve upon two practices as it continues to assess and revise measures in the future. First, while BJA plans to assess the reliability of the new measures after the first quarter of grantees' reporting, officials have not documented, as suggested by best practices, how it will determine if the measures were successful or whether changes would be needed. Second, should future changes to the measures be warranted, BJA could improve the way it documents its decisions and incorporates feedback from stakeholders, including grantees, by recording key methods and assumptions used to guide its revision efforts. By better adhering to best practices identified by GAO and academic literature, BJA could better ensure that its future revision efforts result in successful and reliable metrics-and that the revision steps it has taken are transparent. In the evaluations that GAO reviewed, drug-court program participation was generally associated with lower recidivism. GAO's analysis of evaluations reporting recidivism data for 32 programs showed that drug-court program participants were generally less likely to be re arrested than comparison group members drawn from criminal court, with differences in likelihood reported to be statistically significant for 18 of the programs. Cost-benefit analyses showed mixed results. For example: Across studies showing re-arrest differences, the percentages of drug- court program participants re-arrested were lower than for comparison group members by 6 to 26 percentage points. Drug court participants who completed their program had re-arrest rates 12 to 58 percentage points below those of the comparison group. GAO's analysis of evaluations reporting relapse data for eight programs showed that drug-court program participants were less likely than comparison group members to use drugs, based on drug tests or self- reported drug use, although the difference was not always significant. Of the studies assessing drug-court costs and benefits, the net benefit ranged from positive $47,852 to negative $7,108 per participant.