This paper identifies determinants of extensive (i.e., number of firms) and intensive (i.e., average exports per firm) trade margins, using state-level exports to 187 countries. A standard negative binomial count model is used to handle the issues of non-trading pairs and overdispersion in the extensive trade estimations. In addition to the standard gravity variables of distance and destination-country size, other factors affecting trade costs and export demand are explored. Following directly from a trade model based on heterogeneous firms, these other factors exhibit more consistent and statistically significant effects on the extensive than on the intensive trade margin. One noteworthy finding is that U.S. foreign direct investment has a positive effect on both margins. Also, a given determinant may affect both margins, but not necessarily in the same way. For example, communications infrastructure in the importing country affects the extensive margin positively and the intensive margin negatively.