The offence-related behavior of a total group (n= 25) of MAPPA prisoners in one prison, due for release in the following year to two adjacent probation trust areas, was monitored. Their behaviors in the community were followed up for 1 year. A comparison group (n= 36) was formed of the total number of MAPPA prisoners released from prisons nationally to the same two probation trusts. The frequencies of ADViSOR negative behaviors in prison and the community were strongly correlated, rs (25) = .55, p= .004, as were positive behaviors, rs (25) = .56, p= .004. No statistically significant correlations were found either under usual MAPPA processes in the ADViSOR prison or comparison group prisons. The frequency of ADViSOR negative behaviors statistically significantly predicted, with 92 percent accuracy, the offenders who would reoffend or be recalled to prison (n= 8). Statistically significant similarities in types of behavior were also identified. Results are discussed in terms of the contribution of behavioral monitoring to risk prediction with high-risk offenders, consistency of cross-situational behaviors, and implications for policy and practice. Abstract published by arrangement with John Wiley & Sons.