We iteratively develop and test a model to clarify the relationship between both high and low levels of social (influence) and personal (autonomy) power. A meta-analysis synthesizing primary data (n = 298) and secondary data (n = 498) found that impaired personal power coincided with impaired social power, but not vice versa. Unexpectedly, elevated social power did not coincide with elevated personal power, suggesting that the association between influence and autonomy attenuates with increasing levels of power. Predictions arising from the meta-analysis and our revised theoretical model were supported in a subsequent study (n = 266). We discuss implications of these findings and avenues for future research.