Trade economists usually think about growth in manufactured exports as coming from established firms that diversify into foreign markets and/or increase the share of their output they ship abroad. But this pattern doesn't describe Bangladeshi experiences, where new exporters have typically been new firms that were born to export. These firms have survived in export markets at relatively high rates, and most have sold all of their output abroad. The first portion of this paper documents these patterns, shows that similar but less-striking patterns appear in Chinese data, and finds the patterns to be missing in Taiwan and Colombia. The remainder of the paper interprets these patterns. Specifically, it argues that they are not a consequence of the fact that export processing zones require firms to specialize in foreign sales (most Bangladeshi exports do not originate in EPZs). Rather the distinctive features of Bangladesh's exporting experiences can be explained by the fact that its exports have come from ``orphan industries" with very limited domestic markets. Thus, when profitable exporting opportunities have arisen, entrepreneurs have been unable to exploit them by simply re-directing existing productive capacity toward foreign customers. Rather, they have needed to create new establishments. Using a variant of the Eaton, Eslava, Jinkins, Krizan, Kugler and Tybout (2012) model, we show precisely how start-up costs can influence exporting patterns. First, when entrepreneurs must create productive capacity in order to export, only those producers who expect to sustain large export volumes are likely to enter---that is, sunk entry costs make Melitzian selection effects relatively strong. Second, new exporters are relatively likely to survive in foreign markets. This hysteresis effect obtains because firms in orphan industries cannot reorient their production to domestic consumers when they experience negative shocks to their export profits, nor can they completely recoup their investment in productive capacity by shutting down. Third, it can take an exceptionally large market-wide shock to expected exporting profits before there is much of an export response. But once such a shock has occurred, rapid export growth may follow. This last result obtains partly because potential exporters face similar entry hurdles and, without any domestic market experience, they hold similar expectations about the scope of the market for their products. Thus they are likely to enter in large numbers, if at all. It also reflects the fact that once orphan industry exporters appear, they tend to survive, devoting their entire productive capacity to foreign sales.