We estimate output gaps using three methods for Mainland China on annual data spanning 1982-2003. The estimates are similar and appear to co-move with inflation. Standard Phillips curves, however, do not fit the data well. This may reflect the omission of some important variable(s) such as the effect of price deregulation, trade liberalisation and/or changes in the exchange rate regime. We re-estimate the Phillips curves assuming that there is an unobserved variable that follows an AR(2) process. The modified model fits the data much better and accounts for some of the surprising features of the simple Phillips curve estimates.