Motivated by the argument that managers will cut their dividend only when there are visible signs of poor performance, I revisit the issue of why firms cut their regular cash dividend. I use a propensity score matching methodology to differentiate firms according to their likelihood of cutting a dividend where the likelihood is a function of observable firm characteristics. I have three main findings: First, I find that the market reaction to dividend cut announcements is proportional to the element of surprise. Specifically, for a given magnitude of the dividend cut, I find that the three-day cumulative abnormal return around the dividend cut announcement is more negative for firms with less visible signs of poor performance compared to those that have experienced a more prolonged period of poor performance. Second, while on average firms cut their dividend as a last resort response to poor performance as suggested by prior studies, a significant number of firms cut their dividend pro-actively even without such visible signs of poor performance. The preservation of a low leverage ratio appears to be of first-order importance to these "pro-active" firms. Third, I find that by and large, firms use their poor performance to justify a dividend cut. Moreover, the absence of concurrent poor performance seems to preclude the option of cutting the dividend. Instead, firms may resort to cut back on capital expenditures.