A new map-validation procedure is based on the correlation-coefficient agre
ement between the observed structure-factor magnitudes and their extrapolat
ed values from suitably modified electron-density maps from which they have
been each in turn systematically excluded. The correlation coefficient ten
ds to a maximum as the phase errors in a map are reduced. This principle wa
s used to resolve the single-wavelength anomalous scattering (SAS) and sing
le-derivative isomorphous replacement (SIR) phase ambiguity for a number of
error-free trial structures. Applications employing real data sets tend to
be more difficult owing to data incompleteness and errors affecting the co
nstruction of the Argand diagram.