Line extensions – variants of existing products with new appearances, functions, or forms – constitute a significant fraction of products launched each year. While line extensions typically share components with existing products and, therefore, require lower development costs (as compared to new products), they can also cannibalize the demand of existing products. Thus, the choice of an appropriate set of line extensions is an important issue for a firm. In a seminal paper, Ramdas and Sawhney (2001) model this problem by incorporating both the supply- and demand-side implications of line extensions. We build on this work by further investigating the operational issue of selecting a subset of line extensions from a potential set, with the aim of maximizing profit. Our results for the basic model of Ramdas and Sawhney (2001) include a polynomially-solvable special case and an effective heuristic for the general version. Motivated by practical considerations, we analyze two variants of the basic model: one in which the cardinality of the line extensions is constrained, and the other in which a limited budget is available. We derive a variety of interesting insights on the inherent tradeoffs in the selection of an optimal set of extensions. A key driver of the optimal fraction of extensions is the ratio of the component-specific development cost to the labor cost. The optimal fraction of extensions is a concave and decreasing function of this ratio. Another critical parameter is the density of the product-component bipartite graph – the optimal fraction is an increasing function of this density. We also compare two common discount policies for the material cost of producing the extensions: an all-unit quantity discount and a marginal quantity discount. The marginal discount (resp., all-unit discount) is better when (i) the density of the product-component graph is low (resp., high) or (ii) the density of the product-component graph is modest and there is a large (resp., small) difference between the undiscounted, per-unit prices of the two policies. The percentage markup (i.e., ratio of profit to cost) of the optimal solution is a concave function of the upper bound on the cardinality of the selected line extensions (resp., available budget) and a decreasing function of the ratio of the component-specific development cost to the labor cost. On a carefully designed survey instrument, we observe that (i) only about half of the decision makers demonstrated the correct intuition about the role of product development cost and density of the product-component graph, (ii) almost no one correctly predicted the impact of the discount strategy. Thus, our analysis and results can provide a strong, rigorous foundation for managerial decisions.