Previous analysis of the relationship between mean heat flow q¿ and reduced heat flow q and reduced heat flow qr indicated that qr=0.6q, suggesting that the upper crustal radiogenic contribution to heat flow was coupled to changes in qr. Statistical reexamination of this relationship reveals that this empirical relationship results for several reasons and that other interpretations may apply in stable tectonic areas. Importantly, the slope of the q-qr line changes significantly when certain points are excluded from the regression with the strongest effect exerted by the Basin and Range and eastern Australia. Because these areas were tectonically active during the Cenozoic, they may be outliers, exhibiting disproportionate influence on the regression results. Because q and qr exhibit large errors and are correlated, we used to computer simulation method to fit a straight line to correlated variables with errors. We find that the slope of the line is reduced and that the fit is sensitive to outliers. If we restrict the analysis to more stable tectonic regions, we determine a q-qr relationship with a small slope and large intercept. Inclusion of data from regions which have been tectonically active in the Cenozoic gives results closer to the qr=0.6q relationship. We suggest that this previous result reflects the presence of data from thermally perturbed regions in the q-qr data set. This, in conjunction with the nearly horizontal line found for stable regimes, indicates that coupling between the upper crust and mantle-lower crust system is not imcrust and mantle-lower crust system is not implied. Furthermore, qr for stable regions is about 27 mW/m2 with a small standard deviation (4 mW/m2). The average upper crustal radiogenic component is nearly equivalent but has a large standard deviation (14 mW/m2), suggesting that variations in q for heat flow provinces in stable regions is largely controlled by upper crustal radiogenic variations.