Feedback theory is much less popular now than 5 years ago. However, there is little question that the problem of achieving desired system tolerances from uncertain plants, at minimum cost of feedback, will remain an important, enduring one for many future generations. Although much progress has been made, it is minuscule in comparison with the extent of the problem. The purpose here is to suggest some significant QFT research problems, some tantalizingly on the boundary of the unknown. There have been in the past many suggestions for improvements in feedback synthesis. Most e.g. the Smith Regulator (Int. J. Control 1983;38:977) have been illusory, because they were formulated in a qualitative context, without the disciplines of quantitative uncertainty and performance specifications, degrees of freedom, sensor noise, plant modification, etc. Without such disciplines, it is impossible to properly evaluate competing techniques. The reader is referred to the 1991 Survey paper for some background, Horowitz (Int. J. Control 1991;53(2):255). Copyright (C) 2003 John Wiley Sons, Ltd.