Across the two post-prison employment outcome variables, several patterns of relationships emerged. First, local labor-market conditions, as measured by county unemployment rates, seem to matter by affecting the chances that ex-prisoners find jobs. In the analysis of the initial spell of unemployment, the marginal effect of county unemployment rates on the probability of exiting the initial spell was two percentage points; in the analysis of quarterly employment, the marginal effect reached six percentage points. County unemployment rates equally affected the postprison employment probabilities for offenders with no pre-prison employment and those with some pre-prison employment. Thus, in these Ohio data at least, the speculation that tight labor markets can improve employment opportunities for ex-prisoners finds some support. These findings are tentative, however. Despite the support for the central thesis-that local unemployment rates matter-the unemployment rates observed during much of this period were relatively low. This suggests that the results that were obtained may have come primarily from increases in unemployment leading to job loss, rather than the other way around. If the Ohio prisoners released during 1999 and 2000 benefited from the relatively tight labor-market conditions at the time, which have not been observed since 2000, then this period may represent the maximum gains for ex-prisoner employment that can be obtained from decreases in unemployment. This limitation can be studied by collecting data on post-prison employment experiences for longer periods of time. Once the original investment is made in linking corrections data to UI wage data, the marginal costs of obtaining additional years of data are small. Along these same lines, the use of UI wage data by corrections departments to track employment outcomes seems to be a plausible and relatively inexpensive method for obtaining outcome data on exprisoners. States that collect prisoners' social security numbers, such as Ohio, are in the position to develop relationships with their UI claims departments, and they can enter into computer record matching agreements that permit them to link offender records to the UI wage records. Such arrangements to obtain data can lead to the development of inexpensive, ongoing systems to monitor post-prison outcomes. They also can contribute to the development of research databases. To the extent that several states develop these systems, the various research databases can allow for comparative analyses of post-prison employment outcomes. The speculations about post-prison outcomes also highlight one limitation of this analysis: its reliance only on unemployment rates to mea sure local labor-market conditions. Future research should expand upon the measure of local labor-market conditions to include factors such as the size of the labor force, the sectoral composition, new job growth, wages, and other measures of the supply of and demand for labor. These analyses might give more clues as to whether and how exprisoners can be absorbed when labor markets are not as tight as they were throughout much of this period. Analyses of sectoral demand for labor, for example, can be accomplished by using the industry information in the UI wage records. A second finding is that the accumulation of pre-prison human capital- as indicated by pre-prison employment-seems to facilitate postprison employment. Ex-prisoners with as little as one-quarter of employment in the year prior to admission into prison exited their initial post-prison unemployment more quickly than offenders with no preprison employment during the year prior to admission, and their postprison employment probabilities were as much as 10 percent higher than those with no pre-prison employment. Third, the poor performance of prison programs in this analysis is perplexing and discouraging and merits additional work to better measure program outcomes. The negative outcomes for the vocation certificate recipients can be understood in terms of their role in the prison, the possibility of a spatial mismatch between the trades in which they are trained and local demands for these skills, and the discounting of prison labor by employers. To better understand the program participation outcomes, both better measures of program participation and different methods to isolate causal mechanisms are required. If the results found here were to hold after additional analyses, they would suggest that prison programming does little to enhance the capacity of offenders to compete in local labor markets unless these offenders already possess labor-market skills and experiences. For example, if prison vocation programs focus only on offenders who have prior work experiences, they not only "cream off" the offenders who are most likely to succeed but also fail to serve the majority of offenders who have little to no pre-prison labor-market experience. The combination of the findings about pre-prison employment and education experiences and the absence of findings of beneficial effects of prison program participation also suggest more generally that the current focus on prisoner reentry efforts may be somewhat misplaced, if it focuses too intently on what happens in prison to prepare offenders for release. Without discounting the importance of the need to prepare offenders for reentry, these analyses show that pre-prison work experience and education are much more important in determining postprison outcomes than what goes on in prison to prepare offenders for release. This perspective also suggests that broader labor-market poli cies can have beneficial effects, both for the general labor force and for ex-prisoners. Finally, post-prison supervision also turns out to be positively related to post-prison employment. Whether this arises from offenders' attempting to comply with conditions of supervision or from parole officers' efforts to locate jobs and help in the reintegration effort is not known. Even though states have turned away from using parolerelease decisions, there is still relatively widespread use of different forms of post-prison supervision. The Ohio (and Washington State) cases suggest that continuing to use some form of post-prison supervision can be beneficial in improving post-prison employment. The downside to post-prison supervision is exemplified by California, where more than two-thirds of prison admissions are of persons who have returned to prison due to technical violations of conditions of supervision. While the California case does refute the view that supervision can be beneficial, it suggests that the type and character of supervision may be important. In future research, administrative data such as that used in this paper needs to be supplemented with additional data on post-prison supervision experiences in order to better understand how supervision contributes to post-prison employment and recidivism outcomes.