diff --git a/met/docs/Users_Guide/tc-gen.rst b/met/docs/Users_Guide/tc-gen.rst
index 33b9a73d81..90bf81ec86 100644
--- a/met/docs/Users_Guide/tc-gen.rst
+++ b/met/docs/Users_Guide/tc-gen.rst
@@ -17,7 +17,7 @@ The TC-Gen tool populates a contingency tables with hits, misses, and false alar
 
 Other considerations for interpreting the output of the TC-Gen tool involve the size of the contingency table output. The size of the contingency table will change depending on the number of matches. Additionally, the number of misses is based on the forecast duration and interval (specified in the configuration file). This change is due to the number of model opportunities to forecast the event, which is determined by the specified duration/interval.
 
-Care should be taken when interpreting the statistics for filtered data. In some cases, variables (e.g. storm name) are only available in either the forecast or reference datasets, rather than both. When filtering on a field that is only present in one dataset, the contingency table counts will be impacted. Similarly, the initialization field only impacts the model forecast data. If the valid time (which will impact the reference dataset) isn't also specified, the forecasts will be filtered and matched such that the number of misses will erroneously increase. See section 21.3 for more detail.
+Care should be taken when interpreting the statistics for filtered data. In some cases, variables (e.g. storm name) are only available in either the forecast or reference datasets, rather than both. When filtering on a field that is only present in one dataset, the contingency table counts will be impacted. Similarly, the initialization field only impacts the model forecast data. If the valid time (which will impact the reference dataset) isn't also specified, the forecasts will be filtered and matched such that the number of misses will erroneously increase. See :numref:`tc-gen_practical_info` for more detail.
 
 .. _tc-gen_practical_info: