The object of this study was to validate a technique for establishing inter-rater reliability on the Southwestern Cooperative Interaction Observation Schedule (SCIOS), where it was impractical to bring the observers to a common site. Reliability was originally obtained when eight observers met together. Observers were divided into four pairs. A video tape of a typical classroom scene was transported to each of the pairs in four cities. All observers viewed the same tape within a one-week period. Correlations of each observer with all others were averaged. This average correlation was compared with correlations of observers normally working together. The mean of all the correlations was .457. The corresponding mean for correlations of paired observers was .904. This technique proved to be superior financially and statistically in discriminating need for further training of observers to obtain inter-rater reliability, as compared to using only correlations of paired observers or bringing all observers to a common site. (Author)