The Grand Canonical formalism is generally used in numerical simulatio
ns of finite density QCD since it allows free mobility in the chemical
potential mu. We show that special care has to be used in extracting
numerical results to avoid dramatic rounding effects and spurious tran
sition signals. If we analyze data correctly, with reasonable statisti
cs, no signal of first order phase transition is present and results u
sing the Glasgow prescription are practically coincident with the ones
obtained using the modulus of the fermionic determinant.