This study examined outcomes of an ongoing socialization program for shelter dogs conducted at a local prison. Dogs residing at a Humane Society facility were assigned to either a “Control” or “Socialization” treatment. Dogs assigned to both treatments were administered a pretest at the Humane Society consisting of blood withdrawal for hormone analysis, assessment of responses to commands, and observation of behavior in a novel situation. Dogs assigned to the Socialization treatment were then transported to the prison where they lived with, and were trained by, inmate handlers. Dogs in the Control treatment remained at the Humane Society and received no explicit training. Three weeks later, dogs were administered a posttest identical in form to the pretest. Dogs provided the Socialization, but not Control, treatment exhibited significant improvement from pretest to posttest in compliance with commands. In a novel situation, Socialization dogs showed significantly less jumping on an unfamiliar human and vocalizing, and significantly more yawning, in the posttest relative to the pretest than did Control dogs. Whereas plasma cortisol levels did not vary from pretest to posttest in either group, ACTH levels unexpectedly increased with time in both groups. Moreover, cortisol and ACTH levels were significantly positively correlated with each other at the posttest, but not the pretest. These results provide evidence for positive behavioral outcomes of prison socialization programs for shelter dogs, as well as further support for the notion that shelter housing results in a dysregulation of the hypothalamic–pituitary–adrenal axis.