For all cancer mortality with a ten year lag the estimated ERR for the X-10/Y-12 subcohort was 1.45 per Sv with a likelihood ratio based 95% CI =(0.15, 3.48) - see Table VI and Fig. 2A. Results were also presented for all cancers for two and twenty year lags by facility group - see Table X . The results based on unadjusted doses for the X-10 only subcohort (see line 3 of Table XI ) are consistent with results in earlier studies which showed a positive dose-response relation for all cancers [22,23]. The dissimilarity of dose response associations between X-10, Y-12 and multiple facility workers increases with the lag assumption, as indicated by both the slope estimate and the LRT. A similar pattern is observed for adjusted doses. Results for the twenty year lag are based on doses received before 1965, the time period when radiation monitoring was substantially more complete at X-10 than at Y-12---see [2,3].