The existing literature reports a positive relationship between sacrifice r
atios and central bank independence. This paper discusses two critical poin
ts related to the usual procedures employed by this literature. One concern
s the issue of non-stationarity of the inflation rate, the other the proble
m of correlation between the error terms for different disinflation episode
s in the same country. The way in which the sacrifice ratio-central bank in
dependence relationship is tested is reconsidered taking into account these
points. Our results shed considerable doubts on the nature and robustness
of such a relationship and on the usual interpretation given to it in the l
iterature.