A 1989 study on failure to appear found that information on a defendant’s ties to the community, criminal history, and type and severity of the top arrest charge was able to differentiate between good risks and bad risks. A 2000 study validated these results although changes were observed in the magnitude of the relationship for some of the variables. The model was revised to address policy and practical concerns. A new scale comprised of six items: working telephone in the residence/cellular phone; New York City areas address; expectation of someone at criminal court arraignment; full-time employment, school, or training program; open cases; and prior failure to appear. This proposed recommendation scheme was revalidated on a new and more recent sample of defendants. Data collected in the first quarter of 2001 were used for the validation analysis. The objective was to determine whether it was stable across different samples. Data were drawn from a cohort of arrests made between January 1, 2001, and March 31, 2001, in which defendants were prosecuted on new charges. The dataset contained 91,728 docketed arrests. The results showed that the alternative risk-classification scheme would classify a considerably higher proportion of defendants as low risks, while keeping their failure to appear rate at the same level. Also, the scheme would decrease the proportion of defendants in the moderate- and high-risk categories, while increasing their failure to appear rate. 11 tables, 20 footnotes