diff --git a/NEWS.md b/NEWS.md
index 883783c..85249c9 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,5 +1,5 @@
 # simaerep 0.4.3
-- skip all performance unit tests on CRAN
+- delete performance unit tests (poisson faster than bootstrap) to accommodate CRAN request
 
 # simaerep 0.4.2
 - CRAN submission
diff --git a/cran-comments.md b/cran-comments.md
index 6b3b796..87cc078 100644
--- a/cran-comments.md
+++ b/cran-comments.md
@@ -1,4 +1,16 @@
-# resubmission simaerep 0.4.3
+# resubmission 2 simaerep 0.4.3
+- deleted unit tests which compared function performance when called with different parameters as requested.
+
+Thanks, we see you added
+
+   skip_on_cran() # performance on ci/cd systems is not predictable
+
+but it should not be doing the test anywhere, as WRE explains.
+
+Best,
+Uwe Ligges
+
+# resubmission 1 simaerep 0.4.3
 
 - fix CRAN url in README
 
diff --git a/tests/testthat/test_sim_sites.R b/tests/testthat/test_sim_sites.R
index 47e9821..ef156fd 100644
--- a/tests/testthat/test_sim_sites.R
+++ b/tests/testthat/test_sim_sites.R
@@ -88,16 +88,6 @@ test_that("sim_sites() - prob_low and pval must be between 0 - 1", {
 
 })
 
-test_that("sim_sites() - executing only poisson tests must be faster than using bootstrap simulations", {
-  skip_on_cran() # performance on ci/cd systems is not predictable
-  skip_on_ci()
-
-  t_prob_low <- system.time(sim_sites(df_site_test, df_visit_test, poisson_test = FALSE, prob_lower = TRUE))
-  t_ptest <- system.time(sim_sites(df_site_test, df_visit_test, prob_lower = FALSE, poisson_test = TRUE))
-
-  expect_true(t_prob_low["elapsed"] > t_ptest["elapsed"])
-
-})
 
 test_that("prep_for_sim() - ae vector for site must match number of patients at site", {
 
diff --git a/tests/testthat/test_sim_studies.R b/tests/testthat/test_sim_studies.R
index 4b6f14b..6375c26 100644
--- a/tests/testthat/test_sim_studies.R
+++ b/tests/testthat/test_sim_studies.R
@@ -18,36 +18,6 @@ test_that("sim_studies() - returned dataframe expect no NA and prob_low/pval 0 -
 
 })
 
-test_that("sim_studies() - executing only poisson tests must be faster than using bootstrap simulations", {
-  skip_on_cran() # performance on ci/cd systems is not predictable
-  skip_on_ci()
-
-  t_ptest <- system.time({
-    df_sim_study <- sim_studies(
-      df_site = df_site_test,
-      df_visit = df_visit_test,
-      r = 3,
-      parallel = FALSE,
-      poisson_test = TRUE,
-      prob_lower = FALSE
-    )
-  })
-
-  t_prob_low <- system.time({
-    df_sim_study <- sim_studies(
-      df_site = df_site_test,
-      df_visit = df_visit_test,
-      r = 3,
-      parallel = FALSE,
-      poisson_test = FALSE,
-      prob_lower = TRUE
-    )
-  })
-
-  expect_true(t_ptest["elapsed"] < t_prob_low["elapsed"])
-
-})
-
 
 test_that("sim_studies() with keep_ae = TRUE - retain individual site and study AE", {