Traditional assumptions about the nature of science have historically precluded the empirical study of human agency, but conceptual and methodological advances by Howard (1984) and others have made research on agency possible. Following Howard and Conway's (1986) volitional instruction paradigm, the present study examined whether participants could willfully control their cigarette use. Data from 43 smokers indicated that participants did indeed smoke significantly less when instructed to 'try not to smoke;' moreover, the proportion of variability explained by this simple instruction was greater than is typically seen in psychological research. Self-control was not associated with self-efficacy, nor was self-control enhanced by participants' self-prediction. Self-ratings of effort were found to mediate the effects of volitional instructions on smoking behavior, providing further support for an agentic (versus mechanistic) interpretation of volition effects. The results of this study highlight the importance of considering both agentic and non-agentic influences on human behavior.