In many data base performance studies there is a need to estimate the number of records of a file which qualify in a query, as well as the number of blocks of secondary storage which contain these records. In this paper we present a model of data base contents and data placement on devices. We extend a multivariate statistical model that was used for the estimation of record selectivities in [5] to model the distribution of records that qualify in a query among the blocks of secondary storage. Then we show how to obtain estimates of block selectivities and we compare our estimates with the estimates of previous models.