We investigate the relationship between revenue diversification and volatility for nonprofits. Modern portfolio theory suggests that more diversification reduces volatility at the expense of reduced expected revenue. We find that this relationship should not be taken for granted. We use a new empirical measure of volatility that addresses estimation issues of expected revenue, including heteroskedasticity and the omission of the effect of diversification on expected revenue. We also examine the impact on nonprofits of different types of diversification. We find that the effects of diversification on volatility and expected revenue depend on the compositional change in the portfolio. For example, a more diversified portfolio achieved by replacing earned income with donations reduces both volatility and expected revenue, while replacing investment income with donations to achieve an increase in diversification of the same magnitude reduces volatility and increases expected revenue. This suggests other motives for nonprofit organizations to hold investments.