When facing supply uncertainty caused by exogenous factors such as adverse weather conditions, firms diversify their supply sources following the wisdom of ¿not holding all eggs in one basket¿. We study a firm that decides on investment and production levels of two unreliable but substitutable resources. Applying real options thinking, production decisions account for actual supply capabilities whereas investment decisions are made in advance. To model triangular supply and demand correlations, we adapt the concepts of random capacity and stochastic proportional yield while using concordant ordered random variables. Optimal profit decreases monotonically in supply correlation and increases monotonically in supply¿demand correlation. Optimal resource selection, however, depends on the trivariate interplay of supply and demand and responds non-monotonically to changing correlations. Moreover, supply hedges (i.e., excess capacity at alternative sources) can be optimal even if supply resources are perfectly positively correlated. To accommodate changing degrees of correlation, the firm adjusts the lower-margin capacities under random capacity; but under stochastic proportional production capability, it uses either low- or high-margin capacities to create tailored ¿scale hedges¿ (i.e., excess capacity at one source which can partially substitute for diversification).