diff --git a/R/geom-smooth.r b/R/geom-smooth.r
index 6c10e98ba5..a4e224450f 100644
--- a/R/geom-smooth.r
+++ b/R/geom-smooth.r
@@ -78,8 +78,8 @@
 geom_smooth <- function(mapping = NULL, data = NULL,
                         stat = "smooth", position = "identity",
                         ...,
-                        method = "auto",
-                        formula = y ~ x,
+                        method = NULL,
+                        formula = NULL,
                         se = TRUE,
                         na.rm = FALSE,
                         show.legend = NA,
diff --git a/R/stat-smooth.r b/R/stat-smooth.r
index 86e2e9dcab..a1cc84d762 100644
--- a/R/stat-smooth.r
+++ b/R/stat-smooth.r
@@ -1,8 +1,10 @@
-#' @param method Smoothing method (function) to use, accepts either a character vector,
-#'   e.g. `"auto"`, `"lm"`, `"glm"`, `"gam"`, `"loess"` or a function, e.g.
-#'   `MASS::rlm` or `mgcv::gam`, `stats::lm`, or `stats::loess`.
+#' @param method Smoothing method (function) to use, accepts either
+#'   `NULL` or a character vector, e.g. `"lm"`, `"glm"`, `"gam"`, `"loess"`
+#'   or a function, e.g. `MASS::rlm` or `mgcv::gam`, `stats::lm`, or `stats::loess`.
+#'   `"auto"` is also accepted for backwards compatibility.  It is equivalent to
+#'   `NULL`.
 #'
-#'   For `method = "auto"` the smoothing method is chosen based on the
+#'   For `method = NULL` the smoothing method is chosen based on the
 #'   size of the largest group (across all panels). [stats::loess()] is
 #'   used for less than 1,000 observations; otherwise [mgcv::gam()] is
 #'   used with `formula = y ~ s(x, bs = "cs")` with `method = "REML"`. Somewhat anecdotally,
@@ -10,10 +12,12 @@
 #'   so does not work for larger datasets.
 #'
 #'   If you have fewer than 1,000 observations but want to use the same `gam()`
-#'   model that `method = "auto"` would use, then set
+#'   model that `method = NULL` would use, then set
 #'   `method = "gam", formula = y ~ s(x, bs = "cs")`.
 #' @param formula Formula to use in smoothing function, eg. `y ~ x`,
-#'   `y ~ poly(x, 2)`, `y ~ log(x)`
+#'   `y ~ poly(x, 2)`, `y ~ log(x)`. `NULL` by default, in which case
+#'   `method = NULL` implies `formula = y ~ x` when there are fewer than 1,000
+#'   observations and `formula = y ~ s(x, bs = "cs")` otherwise.
 #' @param se Display confidence interval around smooth? (`TRUE` by default, see
 #'   `level` to control.)
 #' @param fullrange Should the fit span the full range of the plot, or just
@@ -37,8 +41,8 @@
 stat_smooth <- function(mapping = NULL, data = NULL,
                         geom = "smooth", position = "identity",
                         ...,
-                        method = "auto",
-                        formula = y ~ x,
+                        method = NULL,
+                        formula = NULL,
                         se = TRUE,
                         n = 80,
                         span = 0.75,
@@ -77,7 +81,8 @@ stat_smooth <- function(mapping = NULL, data = NULL,
 #' @export
 StatSmooth <- ggproto("StatSmooth", Stat,
   setup_params = function(data, params) {
-    if (identical(params$method, "auto")) {
+    msg <- character()
+    if (is.null(params$method) || identical(params$method, "auto")) {
       # Use loess for small datasets, gam with a cubic regression basis for
       # larger. Based on size of the _largest_ group to avoid bad memory
       # behaviour of loess
@@ -87,18 +92,30 @@ StatSmooth <- ggproto("StatSmooth", Stat,
         params$method <- "loess"
       } else {
         params$method <- "gam"
+      }
+      msg <- c(msg, paste0("method = '", params$method, "'"))
+    }
+
+    if (is.null(params$formula)) {
+      if (identical(params$method, "gam")) {
         params$formula <- y ~ s(x, bs = "cs")
+      } else {
+        params$formula <- y ~ x
       }
-      message(
-        "`geom_smooth()` using method = '", params$method,
-        "' and formula '", deparse(params$formula), "'"
-      )
+      msg <- c(msg, paste0("formula '", deparse(params$formula), "'"))
+    }
+    if (identical(params$method, "gam")) {
+      params$method <- mgcv::gam
+    }
+
+    if (length(msg) > 0) {
+      message("`geom_smooth()` using ", paste0(msg, collapse = " and "))
     }
 
     params
   },
 
-  compute_group = function(data, scales, method = "auto", formula = y ~ x,
+  compute_group = function(data, scales, method = NULL, formula = NULL,
                            se = TRUE, n = 80, span = 0.75, fullrange = FALSE,
                            xseq = NULL, level = 0.95, method.args = list(),
                            na.rm = FALSE) {
diff --git a/man/geom_smooth.Rd b/man/geom_smooth.Rd
index c29a0fa0d0..540207b2db 100644
--- a/man/geom_smooth.Rd
+++ b/man/geom_smooth.Rd
@@ -6,11 +6,11 @@
 \title{Smoothed conditional means}
 \usage{
 geom_smooth(mapping = NULL, data = NULL, stat = "smooth",
-  position = "identity", ..., method = "auto", formula = y ~ x,
+  position = "identity", ..., method = NULL, formula = NULL,
   se = TRUE, na.rm = FALSE, show.legend = NA, inherit.aes = TRUE)
 
 stat_smooth(mapping = NULL, data = NULL, geom = "smooth",
-  position = "identity", ..., method = "auto", formula = y ~ x,
+  position = "identity", ..., method = NULL, formula = NULL,
   se = TRUE, n = 80, span = 0.75, fullrange = FALSE,
   level = 0.95, method.args = list(), na.rm = FALSE,
   show.legend = NA, inherit.aes = TRUE)
@@ -44,11 +44,13 @@ often aesthetics, used to set an aesthetic to a fixed value, like
 \code{colour = "red"} or \code{size = 3}. They may also be parameters
 to the paired geom/stat.}
 
-\item{method}{Smoothing method (function) to use, accepts either a character vector,
-e.g. \code{"auto"}, \code{"lm"}, \code{"glm"}, \code{"gam"}, \code{"loess"} or a function, e.g.
-\code{MASS::rlm} or \code{mgcv::gam}, \code{stats::lm}, or \code{stats::loess}.
+\item{method}{Smoothing method (function) to use, accepts either
+\code{NULL} or a character vector, e.g. \code{"lm"}, \code{"glm"}, \code{"gam"}, \code{"loess"}
+or a function, e.g. \code{MASS::rlm} or \code{mgcv::gam}, \code{stats::lm}, or \code{stats::loess}.
+\code{"auto"} is also accepted for backwards compatibility.  It is equivalent to
+\code{NULL}.
 
-For \code{method = "auto"} the smoothing method is chosen based on the
+For \code{method = NULL} the smoothing method is chosen based on the
 size of the largest group (across all panels). \code{\link[stats:loess]{stats::loess()}} is
 used for less than 1,000 observations; otherwise \code{\link[mgcv:gam]{mgcv::gam()}} is
 used with \code{formula = y ~ s(x, bs = "cs")} with \code{method = "REML"}. Somewhat anecdotally,
@@ -56,11 +58,13 @@ used with \code{formula = y ~ s(x, bs = "cs")} with \code{method = "REML"}. Some
 so does not work for larger datasets.
 
 If you have fewer than 1,000 observations but want to use the same \code{gam()}
-model that \code{method = "auto"} would use, then set
+model that \code{method = NULL} would use, then set
 \code{method = "gam", formula = y ~ s(x, bs = "cs")}.}
 
 \item{formula}{Formula to use in smoothing function, eg. \code{y ~ x},
-\code{y ~ poly(x, 2)}, \code{y ~ log(x)}}
+\code{y ~ poly(x, 2)}, \code{y ~ log(x)}. \code{NULL} by default, in which case
+\code{method = NULL} implies \code{formula = y ~ x} when there are fewer than 1,000
+observations and \code{formula = y ~ s(x, bs = "cs")} otherwise.}
 
 \item{se}{Display confidence interval around smooth? (\code{TRUE} by default, see
 \code{level} to control.)}
diff --git a/tests/testthat/test-geom-smooth.R b/tests/testthat/test-geom-smooth.R
index 0c378eae31..e84f8dae6a 100644
--- a/tests/testthat/test-geom-smooth.R
+++ b/tests/testthat/test-geom-smooth.R
@@ -48,6 +48,15 @@ test_that("default smoothing methods for small and large data sets work", {
     "method = 'gam' and formula 'y ~ s\\(x, bs = \"cs\"\\)"
   )
   expect_equal(plot_data$y, as.numeric(out))
+
+  # backwards compatibility of method = "auto"
+  p <- ggplot(df, aes(x, y)) + geom_smooth(method = "auto")
+
+  expect_message(
+    plot_data <- layer_data(p),
+    "method = 'gam' and formula 'y ~ s\\(x, bs = \"cs\"\\)"
+  )
+  expect_equal(plot_data$y, as.numeric(out))
 })