Tabulated Student'st-values and climatic insensitivity among inner tree-ring widths can bias estimates of statistical significance for cross correlations relating "floating" and master tree-ring series. These biases can be removed by (i) directly computing significance levels for cross-correlation coefficients at dating positions and (ii) deleting insensitive inner rings from a dated floating sample before final correlation analysis. The number of early rings to delete can be determined from plots of cross-correlation coefficients linking a dated floating series of artificially decreasing length with a master series. These modifications improve the precision of Yamaguchi and Allen's approach (D.K. Yamaguchi and G.L. Allen. 1992. Can. J. For. Res. 22: 1215–1221) for estimating significance.