Purpose: The World Health Organization issued its global action plan on physical activities 2018-2030, emphasizing the importance of context-specific evidence on the subject. Accordingly, this study aims to provide unique and important policy insights on trends and drivers of participation in physical exercises by academic community in Sri Lankan universities. Design/methodology/approach: For this purpose, we collected cross-sectional data (n = 456) in 2020 using a survey, and first, estimated a double-hurdle model to uncover covariates influencing likelihood and intensity of physical exercises overall. Second, count-data models are estimated to capture regularity of key exercises. Findings: The results reveal that about 50% of members do not participate in any general physical exercise. Older members (marginal effect (ME) = 3.764, p < 0.01), non-Buddhists (ME = 54.889, p < 0.01) and alcohol consumers (ME = 32.178, p < 0.05) exhibit a higher intensity of participating in exercises overall. The intensity is lower for rural members (ME = -63.807, p < 0.01) and those with health insurance covers (ME = -31.447, p < 0.05). Individuals diagnosed for chronic illnesses show a higher likelihood of exercising but, their time devotion is limited. The number of children the academic staff members have as parents reduces the likelihood, but for those who choose to exercise have higher time devotion with increased number of children. The covariates play a similar role in determining regularity of key exercises: walking, jogging and exercising on workout machines. Research limitations/implications: The results imply a need to promote exercising in general and particularly among younger, healthy, insured and female individuals living in rural sector. Originality/value: The study covers an under-researched professional sub-group in an under-researched developing context, examining both the likelihood and regularity of exercising as both dimensions are equally important for individuals to maintain healthy lives.