diff --git a/mps/code/comm.gmk b/mps/code/comm.gmk
index 0ac2c94256f..e5af99266ae 100644
--- a/mps/code/comm.gmk
+++ b/mps/code/comm.gmk
@@ -327,6 +327,7 @@ MMQA=perl test/qa -i ../code -l ../code/$(PFM)/$(VARIETY)/mps.o
 
 $(PFM)/$(VARIETY)/testmmqa:
 	$(MAKE) -f $(PFM).gmk VARIETY=$(VARIETY) TARGET=mps.o variety
+	(if [ "$(VARIETY)" = "cool" ]; then cd ../test && $(MMQA) runset testsets/coolonly; fi)
 	(cd ../test && $(MMQA) runset testsets/argerr)
 	(cd ../test && $(MMQA) runset testsets/conerr)
 	(cd ../test && $(MMQA) runset testsets/passing)
diff --git a/mps/test/testsets/argerr b/mps/test/testsets/argerr
index 990040192b4..8c587bae4f7 100644
--- a/mps/test/testsets/argerr
+++ b/mps/test/testsets/argerr
@@ -1,3 +1,7 @@
+% This testset contains all the "argerr" test cases that pass in
+% both the cool and hot varieties, together with comments explaining
+% why the other test cases fail.
+
 argerr/0.c
 argerr/1.c
 argerr/2.c
@@ -14,7 +18,7 @@ argerr/12.c
 argerr/13.c
 argerr/14.c
 argerr/15.c
-argerr/16.c
+% argerr/16.c -- assertion in different place in the hot variety
 argerr/17.c
 argerr/18.c
 argerr/19.c
@@ -47,9 +51,9 @@ argerr/45.c
 argerr/46.c
 argerr/47.c
 argerr/48.c
-argerr/49.c
-argerr/50.c
-argerr/51.c
+% argerr/49.c -- AVERT(Rank) does nothing in hot variety
+% argerr/50.c -- AVERT(Rank) does nothing in hot variety
+% argerr/51.c -- AVERT(Rank) does nothing in hot variety
 argerr/52.c
 argerr/53.c
 argerr/54.c
@@ -92,9 +96,9 @@ argerr/90.c
 argerr/91.c
 argerr/92.c
 argerr/93.c
-argerr/94.c
-argerr/95.c
-argerr/96.c
+% argerr/94.c -- AVERT(Rank) does nothing in hot variety
+% argerr/95.c -- AVERT(Rank) does nothing in hot variety
+% argerr/96.c -- AVERT(Rank) does nothing in hot variety
 argerr/97.c
 argerr/98.c
 argerr/99.c
@@ -102,12 +106,12 @@ argerr/100.c
 argerr/101.c
 argerr/102.c
 argerr/103.c
-argerr/104.c
-argerr/105.c
-argerr/106.c
-argerr/107.c
-argerr/108.c
-argerr/109.c
+% argerr/104.c -- AVERT(Rank) does nothing in hot variety
+% argerr/105.c -- AVERT(Rank) does nothing in hot variety
+% argerr/106.c -- AVERT(Rank) does nothing in hot variety
+% argerr/107.c -- AVERT(RootMode) does nothing in hot variety
+% argerr/108.c -- AVERT(RootMode) does nothing in hot variety
+% argerr/109.c -- AVERT(RootMode) does nothing in hot variety
 argerr/110.c
 argerr/111.c
 argerr/112.c
@@ -117,12 +121,12 @@ argerr/115.c
 argerr/116.c
 argerr/117.c
 argerr/118.c
-argerr/119.c
-argerr/120.c
-argerr/121.c
-argerr/122.c
-argerr/123.c
-argerr/124.c
+% argerr/119.c -- AVERT(Rank) does nothing in hot variety
+% argerr/120.c -- AVERT(Rank) does nothing in hot variety
+% argerr/121.c -- AVERT(Rank) does nothing in hot variety
+% argerr/122.c -- AVERT(RootMode) does nothing in hot variety
+% argerr/123.c -- AVERT(RootMode) does nothing in hot variety
+% argerr/124.c -- AVERT(RootMode) does nothing in hot variety
 argerr/125.c
 % argerr/126.c -- see <code/root.c#.fmt.no-align-check>
 argerr/127.c
diff --git a/mps/test/testsets/conerr b/mps/test/testsets/conerr
index 12966b681a0..1bf93cd0712 100644
--- a/mps/test/testsets/conerr
+++ b/mps/test/testsets/conerr
@@ -1,3 +1,7 @@
+% This testset contains all the "argerr" test cases that pass in
+% both the cool and hot varieties, together with comments explaining
+% why the other test cases fail.
+
 conerr/0.c
 conerr/1.c
 conerr/2.c
@@ -15,13 +19,13 @@ conerr/13.c
 conerr/14.c
 conerr/15.c
 conerr/16.c
-conerr/17.c
+% conerr/17.c -- fails in hot variety (assertion is on the critical path)
 conerr/18.c
 conerr/19.c
 conerr/20.c
 conerr/21.c
-conerr/22.c
-conerr/23.c
+% conerr/22.c -- segfaults in hot variety (assertion is on the critical path)
+% conerr/23.c -- segfaults in hot variety (assertion is on the critical path)
 conerr/24.c
 conerr/25.c
 conerr/26.c
diff --git a/mps/test/testsets/coolonly b/mps/test/testsets/coolonly
new file mode 100644
index 00000000000..3125dcbedb5
--- /dev/null
+++ b/mps/test/testsets/coolonly
@@ -0,0 +1,34 @@
+% This testset contains all the test cases that pass only in the cool
+% variety, together with comments explaining why they fail in the hot
+% variety.
+
+% Assertion in different place in the hot variety.
+argerr/16.c
+
+% Rank is not a structure type, so AVERT(Rank) does nothing. 
+argerr/49.c
+argerr/50.c
+argerr/51.c
+argerr/94.c
+argerr/95.c
+argerr/96.c
+argerr/104.c
+argerr/105.c
+argerr/106.c
+argerr/119.c
+argerr/120.c
+argerr/121.c
+
+% RootMode is not a structure type, so AVERT(Rootmode) does nothing. 
+argerr/107.c
+argerr/108.c
+argerr/109.c
+argerr/122.c
+argerr/123.c
+argerr/124.c
+
+% Assertion is on the critical path.
+conerr/17.c
+conerr/22.c
+conerr/23.c
+function/72.c
diff --git a/mps/test/testsets/passing b/mps/test/testsets/passing
index c6893a54c30..85f880b07ce 100644
--- a/mps/test/testsets/passing
+++ b/mps/test/testsets/passing
@@ -1,5 +1,6 @@
-% Test status on OS X
-% $Id$
+% This testset contains all the "function" test cases that pass in
+% both the cool and hot varieties, together with comments explaining
+% why the other test cases fail.
 
 function/0.c
 function/1.c
@@ -72,7 +73,7 @@ function/67.c
 function/69.c
 function/70.c
 function/71.c
-function/72.c
+% function/72.c -- fails in hot variety (assertion is on the critical path)
 function/73.c
 function/74.c
 function/75.c