[en] This study was initially conceived to determine J-R curves from various sizes of specimens to investigate data extrapolation from small size specimens. Instead, this study resulted in the finding of a significant size effect or dependence for the low toughness A 302-B plate investigated. In particular the magnitude of this size dependence is unprecedented for reactor pressure vessel steels. The observed size dependence results in vastly reduced J-R curve toughness levels with increased specimen size, for compact specimens ranging in thickness from 12.7 to 152.4 mm, with all specimens proportional in terms of dimensions. The plate used in this study was specially made to duplicate early production A 302-B plates, which typically exhibit low Charpy-V upper shelf energy levels. The minimal cross-rolling applied to the plate and the high sulfur content result in a high proportion of manganese-sulfide inclusions. The resultant microstructure is one possible explanation for the unexpected results for this plate. Other causes and ideas for future work are also described. An additional observation is that initial crack length-to-width ratio (a/W) did have an influence on the J-R curves for this plate. Specifically, a long crack length (a/W /approximately/ 0.6) can give somewhat higher J-R curve levels than a short crack length (a/W /approximately/ 0.5), for large Δa levels. This finding has a direct impact on the applicability of J/sub D/ and J/sub M/, and data extrapolation. 23 refs., 91 figs., 12 tabs