diff --git a/R/plot.gg_variable.R b/R/plot.gg_variable.R
index e6d7ae38..4e0cf715 100644
--- a/R/plot.gg_variable.R
+++ b/R/plot.gg_variable.R
@@ -526,7 +526,11 @@ plot.gg_variable <- function(x, # nolint: cyclocomp_linter
                 )
             }
           } else {
-            # Factor predictor: jitter + boxplot coloured by observed class
+            # Factor predictor: jitter + boxplot coloured by observed class.
+            # smooth=TRUE is intentionally a no-op here: geom_smooth requires
+            # a continuous x-axis and has no meaningful interpretation for
+            # discrete factor levels.  The boxplot IQR serves as the spread
+            # summary.
             gg_plt[[ind]] <- gg_plt[[ind]] +
               ggplot2::geom_jitter(
                 ggplot2::aes(
@@ -565,6 +569,10 @@ plot.gg_variable <- function(x, # nolint: cyclocomp_linter
                 )
             }
           } else {
+            # Factor predictor (multi-class): boxplot + jitter per facet.
+            # smooth=TRUE is intentionally a no-op here for the same reason
+            # as the binary factor path above — geom_smooth requires a
+            # continuous x-axis.
             gg_plt[[ind]] <- gg_plt[[ind]] +
               ggplot2::geom_boxplot(
                 ggplot2::aes(x = .data$var, y = .data$yhat),
@@ -605,7 +613,10 @@ plot.gg_variable <- function(x, # nolint: cyclocomp_linter
               ggplot2::geom_smooth(ggplot2::aes(x = .data$var, y = .data$yhat), ...)
           }
         } else {
-          # Factor predictor: boxplot + jitter
+          # Factor predictor (regression): boxplot + jitter.
+          # smooth=TRUE is intentionally a no-op here: geom_smooth requires a
+          # continuous x-axis and has no meaningful interpretation for discrete
+          # factor levels.  The boxplot IQR serves as the spread summary.
           gg_plt[[ind]] <- gg_plt[[ind]] +
             ggplot2::geom_boxplot(
               ggplot2::aes(x = .data$var, y = .data$yhat),
diff --git a/tests/testthat/test_gg_variable.R b/tests/testthat/test_gg_variable.R
index e4bd9e40..cbe54087 100644
--- a/tests/testthat/test_gg_variable.R
+++ b/tests/testthat/test_gg_variable.R
@@ -492,3 +492,56 @@ test_that("gg_variable.randomForest: oob=FALSE triggers a warning", {
   expect_s3_class(gg, "gg_variable")
   expect_true("yhat.setosa" %in% names(gg))
 })
+
+## ── smooth=TRUE is a no-op for factor predictors (all families) ──────────────
+
+# geom_smooth requires a continuous x-axis, so smooth=TRUE has no meaning for a
+# factor predictor.  The contract these tests lock in: no GeomSmooth layer is
+# added for a factor x.  Assert that directly (rather than a brittle layer
+# count) so benign plot-composition changes do not break the suite.
+has_smooth_layer <- function(p) {
+  any(vapply(p$layers, function(l) inherits(l$geom, "GeomSmooth"), logical(1)))
+}
+
+test_that("plot.gg_variable binary classification + factor predictor: smooth=TRUE adds no smooth layer", {
+  skip_if_not_installed("randomForest")
+  set.seed(42L)
+  bin_data         <- iris[iris$Species != "virginica", ]
+  bin_data$Species <- droplevels(bin_data$Species)
+  bin_data$size    <- cut(bin_data$Petal.Length, 2L, labels = c("small", "large"))
+  rf  <- randomForest::randomForest(Species ~ size + Sepal.Width, data = bin_data,
+                                    ntree = 25L)
+  gg  <- gg_variable(rf)
+  # smooth=TRUE with a factor predictor must not error
+  expect_no_error(p <- plot(gg, xvar = "size", smooth = TRUE))
+  expect_s3_class(p, "ggplot")
+  expect_false(has_smooth_layer(p))
+})
+
+test_that("plot.gg_variable multi-class classification + factor predictor: smooth=TRUE adds no smooth layer", {
+  skip_if_not_installed("randomForest")
+  set.seed(42L)
+  iris2      <- iris
+  iris2$size <- cut(iris2$Petal.Length, 3L, labels = c("small", "medium", "large"))
+  rf <- randomForest::randomForest(Species ~ size + Sepal.Width, data = iris2,
+                                   ntree = 25L)
+  gg <- gg_variable(rf)
+  # smooth=TRUE with a factor predictor must not error
+  expect_no_error(p <- plot(gg, xvar = "size", smooth = TRUE))
+  expect_s3_class(p, "ggplot")
+  expect_false(has_smooth_layer(p))
+})
+
+test_that("plot.gg_variable regression + factor predictor: smooth=TRUE adds no smooth layer", {
+  skip_if_not_installed("randomForest")
+  set.seed(42L)
+  iris2      <- iris
+  iris2$size <- cut(iris2$Petal.Length, 3L, labels = c("small", "medium", "large"))
+  rf <- randomForest::randomForest(Sepal.Length ~ size + Sepal.Width, data = iris2,
+                                   ntree = 25L)
+  gg <- gg_variable(rf)
+  # smooth=TRUE with a factor predictor must not error
+  expect_no_error(p <- plot(gg, xvar = "size", smooth = TRUE))
+  expect_s3_class(p, "ggplot")
+  expect_false(has_smooth_layer(p))
+})