Individual variability is evident in behavior and physiology of animals. Determining whether behavior at intake may predict subsequent illness in the animal shelter may influence the management of dogs housed at animal shelters and reduce overall disease. While normally associated with mild disease and low mortality rates, respiratory disease nevertheless poses significant challenges to the management of dogs in the stressful environment of animal shelters due to its highly infectious nature. Therefore, the aim of the study was to explore whether behavior at intake can predict subsequent occurrence and progression of upper respiratory disease in dogs at animal shelters. In a correlational study, 84 dogs were assessed throughout their stay at a city animal shelter. The dogs were subjected to a behavioral assessment, 1 min in-kennel behavioral observations across two observation periods, and the collection of urinary cortisol:creatinine (C:C) ratio. The occurrence and progression of upper respiratory disease was monitored through repeated clinical exams (rectal temperature and the occurrence of nasal and ocular discharge, and presence of coughing and sneezing). A basic PLS Path regression model revealed that time in the shelter (estimate = .53, p < .001), and sociability (estimate = .24, p < .001) and curiosity scores (estimate = .09, p = .026) were associated with increased illness. Activity and anxiety scores, however, were not associated with illness. Urinary C:C, taken on the first full day, did not predict subsequent illness when accounting for time. Limitations included attrition of dogs, a small percentage receiving vaccinations, and continuous and non-systematic rotation of dogs in the kennels. Understanding if behavior can predict subsequent illness may improve shelter management practices, and in turn, result in improved live-release outcomes.