Captive welfare studies in odontocete species have been mostly conducted on bottlenose dolphins (Tursiops truncatus) while the welfare of many other species’ -including endangered species- remains poorly studied. More research is needed to find and validate potential indicators of welfare for each species and even for each group. Since captive odontocetes spend most of their time swimming, their swimming features are interesting to study in relation to their welfare state. We first analysed the circular swimming direction bias in three groups of captive odontocetes (Yangtze finless porpoises: Neophocaena asiaeorientalis asiaeorientalis; East-Asian finless porpoises: N. a. sunameri; and bottlenose dolphins, Tursiops truncatus). Second, we studied the effect of environmental and social factors (i.e., time of the day, delay to training, enrichment, potential perturbation, social grouping, public presence and housing pool) on circular swimming, fast swimming, group swimming, synchronous swimming and contact swimming in the three groups. Yangtze finless porpoises exhibited a clockwise swimming bias while East-Asian finless porpoises and bottlenose dolphins swam significantly more in the counter-clockwise direction. Each studied factor significantly impacted the animals’ swimming behaviour slightly differently depending on the group. However, some patterns were common for the three groups: animals seemed to be more active in the morning than at noon and in the afternoon, and enrichment seemed to decrease circular swimming, fast swimming and social swimming (i.e., synchronous, contact and group swimming), while potential perturbations (e.g., pool cleaning, noise) seemed to increase it. In addition, behaviour differed for Yangtze finless porpoises and bottlenose dolphins right before the training or when other animals were being trained, suggesting an anticipation of this event or an excited/frustrated state in this context. Social separation also impacted these animals’ swimming behaviour with less group swimming but more circular swimming, synchronous swimming and fast swimming when separated. The housing pool had an impact on bottlenose dolphins’ behaviour with more circular swimming, more fast swimming and less group swimming when having access to a larger space. The effect of the presence of public was unclear and requires further investigation. From our results, we propose that circular swimming, synchronous swimming and contact swimming could be useful to monitor animals’ emotional state, but that additional parameters should be added (e.g., swimming speed) since these behaviours can be expressed both in quiet and relaxed contexts and in stressful ones. In addition, fast swimming can be a useful indicator of stress for porpoises but might be more ambiguous for bottlenose dolphins that engage in intense social play bouts for instance. Finally, group swimming might be a good behaviour to monitor when wanting to investigate reactions to various conditions or events that can potentially be stressful. We suggest that further research should be conducted on other groups of odontocetes to validate our findings.