Difficulties in developing fully automated AI problem-solving systems have prompted interest in interaction methodologies that support increased user involvement in the solutiongeneration process. This paper considers the merits of interactive problem solvers grounded in the metaphor of advice.taking. An advicetaking system allows users to specify high-level characteristics at runtime of both the desired solution and the problem-solving process to be employed for the task at hand, with the underlying problem-solver applying those directives to guide solution construction. Advisability enables users to influence the behavior and output of problemsolving systems by making recommendations in terms that are natural and meaningful for them. This paper presents a model of advice-taking and characterizes applications for which advisability is best suited. In addition, it describes an ongoing effort to develop an advisable planner that marries an advice-taking interface to state-of-theart planning technology.