The reporting of U(equiv) and its standard uncertainty has a chequered history. In spite of the recommendation of the IUCr Commission on Journals that authors use the definition of U(equiv) of their own choice, possibly without standard uncertainties, there still seems to be some confusion amongst referees and editors about the status of this derived parameter. It is shown that neither of the common definitions are very useful, and that the standard uncertainty computed from the refinement normal matrix is almost worthless. A potential alternative derived parameter is proposed.