This study examined human–animal interactions during the reactivation of a hospital-based therapy dog program during the COVID-19 pandemic. Data were collected from human–dog interactions at an academic medical center in Virginia. Interaction length, participant role, age group (pediatric or adult), and observed gender were recorded. Handler adherence to human and animal safety protocols (donning personal protective equipment (PPE), using hand sanitizer, and limiting visit length) was measured. Observations from 1016 interactions were collected. t-tests and analysis of variance were conducted. Most visit recipients were healthcare workers (71.69%). Patients received longer visits than other participants (F(4880) = 72.90, p = <0.001); post hoc Bonferroni analyses (p = 0.05/4) showed that patients, both adult (M = 2.58 min, SD = 2.24) (95% C.I = 0.35–1.68) and pediatric (M = 5.81, SD = 4.38) (95% C.I. 3.56–4.97), had longer interaction times than healthcare workers (M = 1.56, SD = 1.92) but not visitors (p = 1.00). Gender differences were not statistically significant (t(552) = −0.736), p = 0.462). Hand sanitizer protocols were followed for 80% of interactions. PPE guidelines were followed for 100% of visits. Most interactions occurred with healthcare workers, suggesting that therapy dog visits are needed for this population. High adherence to COVID-19 safety protocols supports the decision to reactivate therapy animal visitation programs in hospitals. Challenges to safety protocol adherence included ultra-brief interactions and crowds of people surrounding the dog/handler teams. Program staff developed a “buddy system” mitigation strategy to minimize departures from safety protocols and reduce canine stress.