Background and Purpose. The Stroke Rehabilitation Assessment of Movement (S
TREAM) is a new clinical measurement tool for evaluating the recovery of vo
luntary movement and basic mobility following stroke. This article presents
the results of 3 substudies examining the reliability (interrater and intr
arater) and internal consistency of STREAM scores. Subjects and Methods. A
"direct-observation reliability study" was conducted on 20 patients who had
strokes and were in a rehabilitation setting. Paris of raters from a group
of 6 participating therapists setting. Pairs of raters from a group of 6 p
articipating therapists provided data to judge interrater agreement. A "vid
eotaped assessments reliability study" was done to assess intrarater and in
terrater agreement on the scoring of videotaped performances using the STRE
AM measure and involved 4 videotaped assessments that were viewed and rated
on 2 occasions by 20 physical therapists. The internal consistency of the
STREAM scores was evaluated for 26 patients who had strokes and who demonst
rated the full range of motor ability. Results. The reliability of the STRE
AM scores was demonstrated by generalizability correlation coefficients of
.99 for total scores and of .96 to .99 for subscale scores. The internal co
nsistency of the STREAM scores was demonstrated by Cronbach alphas of great
er than .98 on the subscales and overall. Conclusion and Discussion. These
high levels of reliability support the use of the STREAM instrument for the
measurement of motor recovery following stroke. Further work on the validi
ty and responsiveness of the STREAM measure is in progress.