The inclusion of life history as a possible influential factor is pivotal in studies on behavior, welfare, and cognition. Shelter dogs have usually experienced a life involving poor social interactions with humans. Thus, we aimed to investigate the behavioral responses of shelter dogs (SDs) and companion dogs (CDs) during the training of two vocal cues (“sit”, “paw”), as well as the possible associations between their responses and the behaviors of trainers. We studied 15 SDs and 15 CDs in up to eight five-minute training sessions. Dogs’ and trainers’ behaviors were recorded and analyzed (through GLM, GLMM, correlation and Mann–Whitney tests). Shelter dogs responded to more cues per session, with shorter latencies and fewer repetitions of cues. Moreover, SDs spent more time wagging their tails. Dogs’ sex and trainers’ behaviors were also associated with differences in dogs’ responses. The use of a reproachful tone of voice was associated with a greater number of cues responded to, shorter latencies, and fewer repetitions of cues. However, this type voice/discourse was also linked to a greater exhibition of non-training behaviors (e.g., exploring the room or jumping on the trainer), and to dogs spending less time next to the trainer and wagging their tails. On the other hand, the use of a neutral tone of voice and laughter, besides being linked to performance, was also associated with longer durations of tail wagging. Furthermore, the duration of the trainers’ orientation to dogs was correlated with the orientation of the dogs to the trainers. Our data suggest that, even when having experienced social deprivation from humans, SDs’ capacities to learn vocal cues were preserved, possibly due to ontogenic homeostasis processes. Shelter dogs’ greater interest in the sessions may be also credited to their socially-deprived routine. Our outcomes also point to an association between friendly interactions during training and dog performance and excitement, which suggests that such interactions may have the potential to improve SD welfare.