The moral dumbfounding phenomenon for harmless taboo violations is often cited as a critical piece of empirical evidence motivating anti-rationalist models of moral judgment and decision-making. Moral dumbfounding purportedly occurs when an individual remains obstinately and steadfastly committed to a moral judgment or decision even after admitting inability to provide reasons and arguments to support it (Haidt, 2001). Early empirical support for the moral dumbfounding phenomenon led some philosophers and psychologists to suggest that affective reactions and intuitions, in contrast with reasons or reasoning, are the predominant drivers of moral judgments and decisions. We investigate an alternative reason-based explanation for moral dumbfounding: that putatively harmless taboo violations are judged to be morally wrong because of the high perceived likelihood that the agents could have caused harm, even though they did not cause harm in actuality. Our results indicate that judgments about the likelihood of causing harm consistently and strongly predicted moral wrongness judgments. Critically, a manipulation drawing attention to harms that could have occurred (but did not actually occur) systematically increased the severity of moral wrongness judgments. Thus, many participants were sensitive to at least one reason — the likelihood of harm — in making their moral judgments about these kinds of taboo violations. We discuss the implications of these findings for rationalist and anti-rationalist models of moral judgment and decision-making.