Tax evasion is a serious issue that influences governmental revenues, IRS enforcement strategies, and tax policy decisions. While audits are the most effective method of enforcing compliance, they are expensive to conduct and the IRS is only able to audit a fraction of the returns filed each year. This suggests that audits alone are not sufficient to curb the billions of dollars of tax evaded by taxpayers each year and that a better understanding of factors influencing compliance decisions is needed to enable policymakers to craft tax policies that maximize voluntary compliance. Prior research tends to model compliance as economic, environmental, or personal decisions; however, this study models it as a multifaceted decision where these three perspective individually and interactively influence compliance. It is the first to decompose perceived detection risk into two dimensions (selection risk and enforcement risk) and investigates how these two dimensions of risk, decision domains (refund or tax due positions), and three personal factors (mental accounting, narcissism, and proactivity) influence taxpayers’ compliance decisions. I conducted a 2x2 fully crossed experiment involving 331 self-employed taxpayers. These taxpayers have opportunities to evade that employed taxpayers do not. For example, they can earn cash income that is not reported to the IRS by third parties. For self-employed taxpayers (especially those wanting to evade), perceived selection and enforcement risks may be distinctly different depending on a taxpayer’s situation, what they believe they can control, and what risk they are willing to accept. For example, selection risk may be perceived as the greatest risk for those with unreported items on their return, while enforcement risk may be more prominent for those perceiving certain levels of selection risk. Thus, I believe self-employed taxpayers are the most appropriate population to sample from and are likely have reasonable variation in the three personal factors of interest. I find that taxpayers do differentiate between selection and enforcement risks but the difference only manifests for taxpayers in certain decision domains. Taxpayers in a refund position (i.e. conservative mindset) had a greater sensitivity to the form of payment (cash vs. check) and appeared to use this information to make inferences about enforcement risk which was significantly different from their perceptions of selection risk. Conversely, tax due taxpayers (i.e. aggressive mindset) appeared to overlook the form of payment and did not assess these two risks as significantly different. Evaluating the full sample suggests that both selection risk and enforcement risk have a positive influence on compliance. Further, these risks interact to influence compliance. Specifically, compliance is greatest when taxpayers perceive a high likelihood of being selected for an audit and enforcement risk only matters when selection risk is low. This finding is interesting and suggests that avoiding interaction with the IRS is a primary objective of taxpayers. In line with my findings of taxpayers perceiving different risks in refund and tax due positions, the influence of risk perceptions on compliance differed for taxpayers in these positions. Refund taxpayers were influenced by both selection and enforcement risk, similar to the full model; however, tax due taxpayers were only influenced by selection risk and appeared to completely overlook enforcement risk when making their reporting decision. Lastly, the study shows that personal characteristics can also influence compliance in the presence of economic and environmental determinants, but some characteristics only manifest in specific decision domains. Of the three personal characteristics investigated, only mental accounting orientation was a significant predictor for the full sample. When the sample was split by decision domain, only proactivity was a predictor of compliance for refund taxpayers, while only mental accounting orientation was a predictor of compliance for due taxpayers. While I did not find results for narcissism and compliance, my subsequent analysis suggests that individual dimensions of narcissism may be better predictors of compliance than the full measure. Specifically, the exploitation dimension was a significant predictor of compliance for those in a tax due position. This study make several contributions to the accounting and tax literatures. First, this study provides support for a two-construct conceptualization for perceived detection risk that includes both selection and enforcement risks. Second, it answers calls to investigate more comprehensive compliance models and finds economic, environmental, and personal characteristics individually and interactively influence compliance. Third, this study investigates three personal factors that have not been investigated in the tax compliance literature. Finally, this study answers calls for research on self-employed taxpayers and suggests that the IRS will be more successful in increasing compliance by playing on taxpayers’ aversion to being selected for an examination than communicating information on the IRS’ ability to detect noncompliance during an examination.