This thesis develops a theory suggesting that the cardiovascular reactivity exhibited during language use is explicable in terms of self-construction processes. Social constructionist ideas regarding the constructive nature of language were drawn on to outline the ways in which individuals obtain and maintain a sense of self in conversations and other episodes of language use. Three factors regarding conversations were identified as central to self-construction processes, namely the context in which the conversation occurs, the content of the language used, and the resources the individual brings to any particular talking episode. This conceptual scheme was then used to interpret and integrate many diverse findings regarding cardiovascular reactivity, resting blood pressure and cardiovascular disease. Based on this theoretical account, it was hypothesized that conversations about the self would be related to greater cardiovascular reactivity than conversations not focused on the self, and further, that conversations about private aspects of oneself would be related to greater cardiovascular reactivity than conversations about public aspects of oneself. The magnitude of differences in reactivity across the three conversations were expected to depend upon various resources the individual brought into the situation, especially their private and public self-consciousness, social competence, tendency to disclose, usual extent of conversations and their usual comfort felt during conversations. To test these hypotheses an experimental procedure was developed where participants had their blood pressure and heart rate monitored every minute (for approximately 35 minutes) by an automatic blood pressure monitor. During this time they were engaged in three conversations with the researcher about private self, public self, and non-self topics. This procedure was subsequently used on 102 women who, following the experiment, completed a questionnaire which included measures of the relevant individual resources. Results showed that as predicted, blood pressure was most reactive when participants talked about aspects of their private self, and least reactive during non-self talk. Heart rate, however, was most reactive when participants talked about aspects of their public self. Of the individual resource variables, usual extent of conversations and usual comfort of conversations modified the differences in reactivity across the private self, public self and non-self talking conditions, both separately and in combination. Differences in diastolic blood pressure and mean arterial pressure reactivity across the three conditions depended on both the usual extent individuals engaged in conversations and how comfortable they usually feel doing so. Unexpectedly, when these resources were considered, reactivity observed during public self talk was significantly different from reactivity observed during either private self or non-self talk. Overall the results broadly supported the present self-construction account of cardiovascular reactivity during language use. They also highlighted the importance of conversational resources, most notably usual extent and comfort of conversations, in affecting cardiovascular reactivity during any specific conversation. The thesis concludes with some reflections on social constructionist ideas, the realist paradigm, and the nature of language in cardiovascular reactivity research.