Background: Yoga may be an ideal early intervention for those with modifiable risk factors for Alzheimer's disease (AD) development. Objective: To examine the effects of Kundalini yoga (KY) training versus memory enhancement training (MET) on the resting-state connectivity of hippocampal subregions in women with subjective memory decline and cardiovascular risk factors for AD. Methods: Participants comprised women with subjective memory decline and cardiovascular risk factors who participated in a parent randomized controlled trial (NCT03503669) of 12-weeks of KY versus MET and completed pre- and post-intervention resting-state magnetic resonance imaging scans (yoga: n = 11, age = 61.45±6.58 years; MET: n = 11, age = 64.55±6.41 years). Group differences in parcellated (Cole-anticevic atlas) hippocampal connectivity changes (post- minus pre-intervention) were evaluated by partial least squares analysis, controlling for age. Correlations between hippocampal connectivity and perceived stress and frequency of forgetting (assessed by questionnaires) were also evaluated. Results: A left anterior hippocampal subregion assigned to the default mode network (DMN) in the Cole-anticevic atlas showed greater increases in connectivity with largely ventral visual stream regions with KY than with MET (p < 0.001), which showed associations with lower stress (p < 0.05). Several posterior hippocampal subregions assigned to sensory-based networks in the Cole-anticevic atlas showed greater increases in connectivity with regions largely in the DMN and frontoparietal network with MET than with KY (p < 0.001), which showed associations with lower frequency of forgetting (p < 0.05). Conclusion: KY training may better target stress-related hippocampal connectivity, whereas MET may better target hippocampal sensory-integration supporting better memory reliability, in women with subjective memory decline and cardiovascular risk factors. [ABSTRACT FROM AUTHOR]