Research undertaken to develop measures of deep understanding of Ameri
can history for secondary school students serves as a method to study
performance assessment in general. The assessment-policy context condi
tions conceptual and technical choices in design, reinforcing the impo
rtance of linking large-scale performance assessment and classroom use
s. Using design perspective constraints such as desired instructional
change, the state of agreement in the field of history and the practic
ality of assessment for wide-scale use are considered. A model of unde
rstanding adapted from Kintsch's (1992) construction-integration theor
y underlies the assessment. Performance differences between expert and
novice explanations of primary historical sources were used as the ba
sis of the development of a general scoring rubric. The task syntax-in
cluding a measure of prior knowledge, literal comprehension of primary
texts, and an extended written explanation-has been applied to six di
fferent history topics with comparable results. Technical research is
briefly summarized, including the development and validation of the sc
oring rubric, and its application to different history topics.