diff --git a/crates/polars-core/src/frame/explode.rs b/crates/polars-core/src/frame/explode.rs
index eb4fe1e7cf6d..38fdfca1e262 100644
--- a/crates/polars-core/src/frame/explode.rs
+++ b/crates/polars-core/src/frame/explode.rs
@@ -34,6 +34,22 @@ pub struct MeltArgs {
     pub streamable: bool,
 }
 
+impl MeltArgs {
+    pub fn is_valid_schema(&self, schema: &Schema) -> PolarsResult<()> {
+        let mut provided = PlHashSet::new();
+        for name in self.id_vars.iter().chain(self.value_vars.iter()) {
+            if !schema.contains(name) {
+                polars_bail!(ColumnNotFound: "{}", name)
+            }
+
+            if !provided.insert(name) {
+                polars_bail!(Duplicate: "column name '{}' provided more than once in melt", name)
+            }
+        }
+        Ok(())
+    }
+}
+
 impl DataFrame {
     pub fn explode_impl(&self, mut columns: Vec<Series>) -> PolarsResult<DataFrame> {
         polars_ensure!(!columns.is_empty(), InvalidOperation: "no columns provided in explode");
@@ -256,6 +272,9 @@ impl DataFrame {
     /// Similar to melt, but without generics. This may be easier if you want to pass
     /// an empty `id_vars` or empty `value_vars`.
     pub fn melt2(&self, args: MeltArgs) -> PolarsResult<Self> {
+        let schema = self.schema();
+        args.is_valid_schema(&schema)?;
+
         let id_vars = args.id_vars;
         let mut value_vars = args.value_vars;
 
@@ -293,7 +312,6 @@ impl DataFrame {
         }
 
         // values will all be placed in single column, so we must find their supertype
-        let schema = self.schema();
         let mut iter = value_vars.iter().map(|v| {
             schema
                 .get(v)
diff --git a/crates/polars-plan/src/logical_plan/builder.rs b/crates/polars-plan/src/logical_plan/builder.rs
index 7129be8534e5..4543b10a9853 100644
--- a/crates/polars-plan/src/logical_plan/builder.rs
+++ b/crates/polars-plan/src/logical_plan/builder.rs
@@ -857,8 +857,9 @@ impl LogicalPlanBuilder {
     }
 
     pub fn melt(self, args: Arc<MeltArgs>) -> Self {
-        let schema = try_delayed!(self.0.schema(), &self.0, into);
-        let schema = det_melt_schema(&args, &schema);
+        let input_schema = try_delayed!(self.0.schema(), &self.0, into);
+        try_delayed!(args.is_valid_schema(&input_schema), &self.0, into);
+        let schema = det_melt_schema(&args, &input_schema);
         LogicalPlan::MapFunction {
             input: Arc::new(self.0),
             function: FunctionNode::Melt { args, schema },
diff --git a/py-polars/tests/unit/operations/test_melt.py b/py-polars/tests/unit/operations/test_melt.py
index 2d75ab480c1f..a2579154482b 100644
--- a/py-polars/tests/unit/operations/test_melt.py
+++ b/py-polars/tests/unit/operations/test_melt.py
@@ -1,3 +1,5 @@
+import pytest
+
 import polars as pl
 import polars.selectors as cs
 from polars.testing import assert_frame_equal
@@ -48,6 +50,34 @@ def test_melt() -> None:
         ]
 
 
+def test_melt_duplicates() -> None:
+    df = pl.DataFrame({"A": ["a", "b", "c"], "B": [1, 3, 5], "C": [2, 4, 6]})
+
+    with pytest.raises(pl.DuplicateError):
+        df.melt(id_vars=["B", "B"])
+    with pytest.raises(pl.DuplicateError):
+        df.lazy().melt(id_vars=["B", "B"]).collect()
+
+    with pytest.raises(pl.DuplicateError):
+        df.melt(id_vars=["A", "B"], value_vars=["C", "A"])
+    with pytest.raises(pl.DuplicateError):
+        df.lazy().melt(id_vars=["A", "B"], value_vars=["C", "A"]).collect()
+
+
+def test_melt_missing_columns() -> None:
+    df = pl.DataFrame({"A": ["a", "b", "c"], "B": [1, 3, 5], "C": [2, 4, 6]})
+
+    with pytest.raises(pl.ColumnNotFoundError):
+        df.melt(id_vars=["A", "D"])
+    with pytest.raises(pl.ColumnNotFoundError):
+        df.lazy().melt(id_vars=["A", "D"]).collect()
+
+    with pytest.raises(pl.ColumnNotFoundError):
+        df.melt(id_vars=["A"], value_vars=["D"])
+    with pytest.raises(pl.ColumnNotFoundError):
+        df.lazy().melt(id_vars=["A"], value_vars=["D"]).collect()
+
+
 def test_melt_projection_pd_7747() -> None:
     df = pl.LazyFrame(
         {
@@ -81,3 +111,14 @@ def test_melt_no_value_vars() -> None:
         schema={"a": pl.Int64, "variable": pl.String, "value": pl.Null}
     )
     assert_frame_equal(result, expected)
+
+
+# https://github.com/pola-rs/polars/issues/13493
+def test_melt_missing_column_13493() -> None:
+    with pytest.raises(pl.ColumnNotFoundError):
+        (
+            pl.DataFrame({"col0": [0], "col1": ["a"], "col2": ["b"]})
+            .lazy()
+            .melt(id_vars="row_nr")
+            .collect()
+        )