Summary. A group of 50 backward readers aged 9 to 10 years were selected from a total age group of 697 pupils in 16 primary schools of a small local education authority. A battery of tests was given to this group and to a control group of 50 pupils of normal reading ability. The test scores were intercorrelated, separately for the two groups, and in each case a Principal Components Analysis was carried out with rotation to a varimax criterion. The factor patterns of the two groups were different. In the backward readers, but not in the normal readers, reading age was clearly linked with test performance, which seems to depend upon the integrity of central processes. Moreover, the study throws some light on the previous literature describing the results of single tests given to backward readers and matched groups of normal readers, which tends to be inconclusive.