The correct measurement of house price movements over time on a regional level and for small market segments, with relatively few transactions, is considered. The research is based on data of housing market transactions in the Amsterdam Region over the period 1985-1999. In response to the ëconstant qualityí price index problem two main methodologies for index construction have been adopted: a Fixed-sample approach and a Hedonic approach. In the hedonic price model a Hierarchical Trend Model is used for estimation. A Kalman filter could be applied to estimate models of this kind. Price index series, of the Hedonic type and the Fixed-sample type, are presented for the region as a whole as well as per segment and house type within the region. The analysis shows that, over the period studied here, the Hedonic index numbers are superior to the Fixed-sample index numbers, being less sensitive to small market segments and small housing categories.