diff --git a/sklearn_extra/cluster/_k_medoids.py b/sklearn_extra/cluster/_k_medoids.py
index f9d964df..f8e23a3e 100644
--- a/sklearn_extra/cluster/_k_medoids.py
+++ b/sklearn_extra/cluster/_k_medoids.py
@@ -155,6 +155,7 @@ def __init__(
         init="heuristic",
         max_iter=300,
         random_state=None,
+        **metrickw
     ):
         self.n_clusters = n_clusters
         self.metric = metric
@@ -162,6 +163,7 @@ def __init__(
         self.init = init
         self.max_iter = max_iter
         self.random_state = random_state
+        self.metrickw=metrickw
 
     def _check_nonnegative_int(self, value, desc, strict=True):
         """Validates if value is a valid integer > 0"""
@@ -235,7 +237,7 @@ def fit(self, X, y=None):
                 % (self.n_clusters, X.shape[0])
             )
 
-        D = pairwise_distances(X, metric=self.metric)
+        D = pairwise_distances(X, metric=self.metric,**self.metrickwm)
 
         medoid_idxs = self._initialize_medoids(
             D, self.n_clusters, random_state_, X
@@ -379,10 +381,10 @@ def transform(self, X):
             check_is_fitted(self, "cluster_centers_")
 
             Y = self.cluster_centers_
-            kwargs = {}
+
             if self.metric == "seuclidean":
                 kwargs["V"] = np.var(np.vstack([X, Y]), axis=0, ddof=1)
-            DXY = pairwise_distances(X, Y=Y, metric=self.metric, **kwargs)
+            DXY = pairwise_distances(X, Y=Y, metric=self.metric, **self.metrickw)
 
             return DXY
 
@@ -421,7 +423,7 @@ def predict(self, X):
                 X,
                 Y=self.cluster_centers_,
                 metric=self.metric,
-                metric_kwargs=kwargs,
+                metric_kwargs=self.metrickw,
             )
 
             return pd_argmin