Previous analysis has shown that inclusion of regional information imp
roves at-site estimation of point rainfalls and makes it possible to o
btain estimates at non-monitored sites. The basis for this analysis wa
s a partial duration series (PDS) modelling of individual rainfall obs
ervations and use of regional prior distributions for the PDS-paramete
rs in a Bayesian estimation procedure. The advantages of this theoreti
cally satisfactory, but also somewhat complicated procedure are evalua
ted by means of a comparison with simplified procedures. These include
modelling based on regional pooling of all point rainfall data into o
ne sample from a common parent distribution and modelling with disrega
rd of either the dependence between stations or the regional heterogen
eity. The different models are analysed and compared with respect to t
he uncertainty of the predicted extreme events.