This article illustrates how test specifications based solely on academic content standards, without attention to other cognitive skills and item response demands, can fall short of their targeted constructs. First, the authors inductively describe the science achievement construct represented by a statewide sixth-grade science proficiency test. Then they describe the science achievement construct intended to be represented by the test, based on the test blueprint. After comparing the two versions--the intended and achieved constructs--they discuss features of achievement constructs defined to guide test design and development. (Contains 2 tables.)