In this paper, we examined both stated norms of gender preference and actual sex-biases in parental investment in a Tibetan pastoralist society. We collected detailed demographic data to examine how biased parental investment had an effect on infant mortality, infant feeding, the length of interbirth intervals and a decision when giving gifts. Our results indicate a mismatch between self-reported son preference and measures of actual parental investment that favour daughters. We interpret this female-biased parental investment as a possible response to daughters generating more economic resources. However, the stated gender preferences of both sexes reflect cultural norms that appear to have remained unchanged over a long period, which may reflect the importance of male roles in the past. Our behavioural measures of parental investment are those most likely to be in the control of women (such as breastfeeding and interbirth interval), so this mismatch between stated preferences and actual biases in investment may be especially true of women.