An attractive power plant candidate must provide power >70% of the time in a given operational year, typically implying that the frequency of key component failures resulting in unplanned loss of plant availability must be reduced to <0.001/yr. Present fusion devices typically have little motivation to operate with such high reliability and allow relatively frequent instability-driven plasma-terminating events known as disruptions. The vision of an operational fusion reactor therefore includes a level of reliable control performance and confidence well beyond that of presently operating devices. Maximizing use of the limited number of discharges planned for ITER also implies a major advance in control reliability. Fortunately, the mature field of control theory offers methods that routinely provide such levels of performance in many fields from aerospace to process control. [first paragraph from extended abstract]