Abstract
The so-called "threshold" theorem says that, once the error rate per qubit per gate is below a certain value, indefinitely long quantum computation becomes feasible, even if all of the qubits involved are subject to relaxation processes, and all the manipulations with qubits are not exact. The purpose of this article, intended for physicists, is to outline the ideas of quantum error correction and to take a look at the proposed technical instruction for fault-tolerant quantum computation. It seems that the mathematics behind the threshold theorem is somewhat detached from the physical reality, and that some ideal elements are always present in the construction. This raises serious doubts about the possibility of large scale quantum computations, even as a matter of principle.