Research Question The 2008 crisis revealed a surprising amount of risk-taking in very profitable financial institutions. This seems to contradict the traditional predictions of corporate finance models where more profitable firms take less risk, because their shareholders stand to lose more if downside risks realize (Keeley, 1990). Understanding why some of the most profitable financial institutions chose to become exposed to risky market-based instruments on a large scale seems critical for both bank risk management and systemic risk regulation. Contribution In traditional corporate finance models, banks choose the risk of a portfolio of a fixed size. We show that bank risk-taking may take a different form when balance sheet size is a bank's choice. In our model, banks are endowed with a stable 'core business' and choose the size and risk of the market-based investments undertaken alongside their core business. Higher profitability of their core business enables banks to borrow more and engage in risky side activities on a larger scale. Results The model offers predictions that are consistent with some notable patterns of banks' precrisis risk-taking. One pattern is that many banks affected by the crisis were in advanced, rather than developing economies. Our model suggests that risk-taking in profitable banks is more likely when bank leverage constraints are looser, consistent with buoyant credit markets and a pre-crisis weakening of bank capital regulation in many advanced economies. Another pattern is the role of senior funding (e.g. repo) in supporting banks' pre-crisis expansion. Consistent with this evidence, our model confirms that banks' risk-taking in side investments is more severe when these can be financed with senior debt.