This paper examines four hypotheses: (i); in Brazil, as in otherperipheral countries in the post-crisis context, a policy choice appears tohave been made for a flexible exchange rate within a currency band ("dirtyfloat");; (ii); the underlying reasons for this policy appear to have more to dowith pass-through of exchange-rate variations and precautionary demandfor reserves than with the maintenance of a competitive real exchangerate; (iii); in the country's peculiar situation, considerable capital mobility isconjoined with large and liquid financial derivatives markets and a reservesbuild-up policy that carries a high fiscal cost; (iv); until April 2006, reservesaccumulated in much the same way under the floating exchange-ratesystem as they had under the currency band regime; there have beenchanges since then owing to the rapid growth of reserves.