diff --git a/xarray/plot/dataarray_plot.py b/xarray/plot/dataarray_plot.py
index e06a0b7187d..9200557a757 100644
--- a/xarray/plot/dataarray_plot.py
+++ b/xarray/plot/dataarray_plot.py
@@ -515,7 +515,43 @@ def line(
 
     _ensure_plottable(xplt_val, yplt_val)
 
-    primitive = ax.plot(xplt_val, yplt_val, *args, **kwargs)
+    # When plotting 2D data with hue, group lines by hue value so that
+    # lines sharing the same hue are drawn with the same color. Without
+    # this, ax.plot draws each column as a separate line with a new color
+    # from the cycle, causing duplicate hue values to get different colors.
+    # The 2D array can be xplt_val or yplt_val depending on orientation.
+    _has_2d_data = xplt_val.ndim == 2 or yplt_val.ndim == 2
+    if hueplt is not None and _has_2d_data:
+        import matplotlib as mpl
+
+        hue_values = hueplt.to_numpy()
+        # Unique values in order of first appearance
+        unique_hues = list(dict.fromkeys(hue_values))
+
+        # Only assign per-hue colors if the caller didn't set one explicitly
+        user_set_color = "color" in kwargs or "c" in kwargs
+        if not user_set_color:
+            prop_cycle = mpl.rcParams["axes.prop_cycle"]
+            cycle_colors = [p["color"] for p in prop_cycle]
+
+        primitive = []
+        legend_handles = []
+        for idx, hue_val in enumerate(unique_hues):
+            plot_kwargs = dict(kwargs)
+            if not user_set_color:
+                plot_kwargs["color"] = cycle_colors[idx % len(cycle_colors)]
+            # Find all columns belonging to this hue group
+            col_indices = [i for i, h in enumerate(hue_values) if h == hue_val]
+            for col_idx in col_indices:
+                x_vals = xplt_val[:, col_idx] if xplt_val.ndim == 2 else xplt_val
+                y_vals = yplt_val[:, col_idx] if yplt_val.ndim == 2 else yplt_val
+                lines = ax.plot(x_vals, y_vals, *args, **plot_kwargs)
+                primitive.extend(lines)
+                # Keep only the first line per group for the legend
+                if col_idx == col_indices[0]:
+                    legend_handles.extend(lines)
+    else:
+        primitive = ax.plot(xplt_val, yplt_val, *args, **kwargs)
 
     if _labels:
         if xlabel is not None:
@@ -528,7 +564,13 @@ def line(
 
     if darray.ndim == 2 and add_legend:
         assert hueplt is not None
-        ax.legend(handles=primitive, labels=list(hueplt.to_numpy()), title=hue_label)
+        hue_values = hueplt.to_numpy()
+        unique_hues = list(dict.fromkeys(hue_values))
+        ax.legend(
+            handles=legend_handles,
+            labels=[str(h) for h in unique_hues],
+            title=hue_label,
+        )
 
     if isinstance(xplt.dtype, np.dtype) and np.issubdtype(xplt.dtype, np.datetime64):  # type: ignore[redundant-expr]
         _set_concise_date(ax, axis="x")