This prospective study examined the associations of dog/cat ownership with incident disabling dementia using propensity score matching based on the physical, social, and psychological characteristics of dog and cat owners. We also examined associations of the interaction between dog/cat ownership and exercise habit and social isolation with dementia. Overall, 11,194 older adults selected using stratified and random sampling strategies in 2016 were analyzed. Dog/cat ownership was defined as “current” or “past and never”. Disabling dementia was defined according to physicians’ rating in the long-term care insurance system in Japan during the approximately 4-year follow-up period. Statistical analysis was weighted by the inverse of the propensity score in the generalized estimating equation after adjusting for follow-up period. Current dog owners (8.6 %) had an odds ratio (OR) of 0.60 (95 %CI: 0.37–0.977) of having disabling dementia compared to past and never owners. For cat ownership, the corresponding OR was 0.98 (95 %CI: 0.62–1.55). Current dog owners with a regular exercise habit had an OR of 0.37 (0.20–0.68) compared to past and never dog owners with no exercise habit. Further, current dog owners with no social isolation had an OR of 0.41 (0.23–0.73) compared to past and never dog owners with social isolation. Dog ownership had a suppressive effect on incident disabling dementia after adjusting for background factors over a 4-year follow-up period. Specifically, dog owners with an exercise habit and no social isolation had a significantly lower risk of disabling dementia.