diff --git a/CHANGELOG.md b/CHANGELOG.md
index 732746df02..83be49db79 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -9,6 +9,7 @@
 * plot_posterior: fix overlap of hdi and rope (#1263)
 * `plot_dist` bins argument error fixed (#1306)
 * improve handling of circular variables in `az.summary` (#1313)
+* Removed change of default warning in `ELPDData` string representation (#1321)
 * update `radon` example dataset to current InferenceData schema specification (#1320)
 
 ### Deprecation
diff --git a/arviz/stats/stats_utils.py b/arviz/stats/stats_utils.py
index a74f1ecefd..24fe53270a 100644
--- a/arviz/stats/stats_utils.py
+++ b/arviz/stats/stats_utils.py
@@ -437,11 +437,6 @@ def get_log_likelihood(idata, var_name=None):
    (0.7, 1]   (bad)      {{4:{0}d}} {{8:6.1f}}%
    (1, Inf)   (very bad) {{5:{0}d}} {{9:6.1f}}%
 """
-SCALE_WARNING_FORMAT = """
-The scale is now log by default. Use 'scale' argument or 'stats.ic_scale' rcParam if
-you rely on a specific value.
-A higher log-score (or a lower deviance) indicates a model with better predictive
-accuracy."""
 SCALE_DICT = {"deviance": "deviance", "log": "elpd", "negative_log": "-elpd"}
 
 
@@ -478,7 +473,6 @@ def __str__(self):
                 "Count", "Pct.", *[*counts, *(counts / np.sum(counts) * 100)]
             )
             base = "\n".join([base, extended])
-        base = "\n".join([base, SCALE_WARNING_FORMAT])
         return base
 
     def __repr__(self):