Optimization approaches, combinatorial and continuous, to a capital-budgeting problem (CBP) are presented. This NP-hard problem, traditionally modelled as a linear binary problem, is represented as a biquadratic over an intersection of a sphere and a upersphere. This allows applying nonlinear optimization to it. Also, the method of combinatorial and surface cuttings (MCSC) is adopted to (CBP). For the single constrained version (1CBP), new combinatorial models are introduced based on joint analysis of the constraint, objective function, and feasible region. Equivalence of (1CBP) to the multichoice knapsack problem (MCKP) is shown. Peculiarities of Branch&Bound techniques to (1CBP) are described.