Bertola/Caballero (1994) and Abel/Eberly (1996) extended Jor- genson.s classical model of .rms.optimal investment. By introducing investment frictions, they were able to capture the role of future antic- ipations in investment decisions as well as the lumpy and intermittent nature of investment dynamics. We extend Jorgenson.s model to the other direction of .nancing frictions. We construct a model of an equity-only .rm, who must pay a linear .nancing cost for issuing new shares. We show that the .rm.s optimal investment-.nancing is a two-trigger policy in which the .rm .nances investment by issuing new shares (supplementing internal funds) when the shadow price of capital hits the upper trigger value. When the shadow price hits the lower trigger value, she sells a portion of her capital stock and buys back shares (or pays dividends). Values of the shadow price of capital between the two trigger values de.ne a range of "inaction", in which the .rm does neither issue nor buy back shares and invests all of her internal funds for expansion.