The paper analyzes the euro-area interbank market. The martingale hypothesis for the Eonia, the reference overnight interest rate, is tested and rejected. Such rejection is a sufficient condition for a liquidity effect, which is then estimated. The magnitude of the effect is found to depend on the perceived degree of persistence of the liquidity shock. At the beginning of the reserve maintenance period a liquidity drain amounting to 3 per cent of required reserves raises the Eonia by 4 basis points, by 13-15 points, by 25 points or more (up to the limits of the official rate corridor, i.e. roughly Â± 100 basis points), depending on whether it is expected to be purely temporary, to last at least through the following day or through the rest of the holding period. Non-purely-temporary effects may take place when the liquidity shock has some signaling value for the monetary policy stance; however, little if any evidence of shocks of this kind is found. The liquidity effect is read off the slope of a euro-area-wide demand equation for daily reserves which incorporates the current as well as the expected overnight rate among the regressors. The two elasticities are very similar in absolute value and have opposite signs; this is consistent with the announcement effect, the ability by the central bank to influence the current rate without resorting to open market operations. The area-wide demand curve is retrieved by estimating separate relationships for each of the 11 euro-area national banking systems. Some heterogeneity across the different countries is detected. In particular, in some cases the demand for reserves turns out to be interest rate-inelastic over the holding period, suggesting that there is room for further efficiency improvements.