This study examined the construct validity of the subtests in the California Entry Level Test (ELT). The ELT is administered to every first grade pupil in California as part of the California Assessment Program and is used as a baseline measure of the prereading skills of beginning first graders. The discriminant validity of the subtests was demonstrated by a factor analysis of item scores from 3,010 pupils (a one per cent random sample of all first graders tested). A Principal components analysis, followed by varimax rotation, yielded a factor structure analogous to the test structure. Items from each of the five subtests loaded only on their own factor except for a few language development items which had secondary loadings with other factors. Multitrait-multimethod correlation matrices were used to determine the agreement of ELT subtests with corresponding subtests in four popular readiness tests. Findings were mixed. Subtests had convergent validity with other measures of the same construct, but only the subtest with the greatest variance had discriminant validity consistently. (Author)