diff --git a/NEWS.md b/NEWS.md
index db94e422..b80ca542 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,5 +1,9 @@
 # reprex (development version)
 
+* `reprex()` no longer includes the full traceback by default, as this is 
+  only useful in relatively rare situations, and otherwise adds a bunch
+  of clutter (#448).
+
 The unexported `prex_*()` functions protect the current session from option changes coming from reprex's own machinery, such as disabling color (#427).
 
 # reprex 2.0.2
diff --git a/R/reprex.R b/R/reprex.R
index db7b9e31..f6920cb1 100644
--- a/R/reprex.R
+++ b/R/reprex.R
@@ -40,16 +40,10 @@
 #'
 #' @section Error backtraces:
 #'
-#' reprex sets the rlang option `rlang_backtrace_on_error_report = "full"`.
-#' Combined with the knitr option `error = TRUE`, this means rlang errors are
-#' displayed with a full backtrace. This basically eliminates the need to call
-#' [rlang::last_error()] or [rlang::last_trace()] explicitly, although these
-#' functions can be used in a reprex.
-#'
-
-#' Insert a line containing the special comment `#'` in between the
-#' error-causing code and the `last_error()` or `last_trace()` call, to fulfill
-#' the requirement of being in separate chunks:
+#' To use [rlang::last_error()] or [rlang::last_trace()] within a reprex,
+#' you must place them in a different "chunk" to the code that generates an
+#' error. The easiest way to do is to insert a line containing the special
+#' comment `#'` after error-causing code:
 
 #' ``` r
 #' f <- function() rlang::abort('foo')
diff --git a/R/reprex_document.R b/R/reprex_document.R
index bbe5a789..7fccbeb8 100644
--- a/R/reprex_document.R
+++ b/R/reprex_document.R
@@ -135,7 +135,6 @@ reprex_opts <- function(venue = "gh") {
     ```{{r reprex-options, include = FALSE}}
     options(
       keep.source = TRUE,
-      rlang_backtrace_on_error_report = "full",
       crayon.enabled = FALSE,
       reprex.current_venue = "{venue}"
     )
diff --git a/R/reprex_render.R b/R/reprex_render.R
index 5b8ae30e..b0e2145c 100644
--- a/R/reprex_render.R
+++ b/R/reprex_render.R
@@ -153,7 +153,6 @@ reprex_render_impl <- function(input,
     # should be kept in sync with what reprex_opts() sets
     opts_to_safeguard <- options(
       "keep.source",
-      "rlang_backtrace_on_error_report",
       "crayon.enabled"
     )
     withr::defer(options(opts_to_safeguard))
diff --git a/man/reprex.Rd b/man/reprex.Rd
index bc2d7ec0..7b71f2b0 100644
--- a/man/reprex.Rd
+++ b/man/reprex.Rd
@@ -163,15 +163,10 @@ code (see examples).
 \section{Error backtraces}{
 
 
-reprex sets the rlang option \code{rlang_backtrace_on_error_report = "full"}.
-Combined with the knitr option \code{error = TRUE}, this means rlang errors are
-displayed with a full backtrace. This basically eliminates the need to call
-\code{\link[rlang:last_error]{rlang::last_error()}} or \code{\link[rlang:last_error]{rlang::last_trace()}} explicitly, although these
-functions can be used in a reprex.
-
-Insert a line containing the special comment \verb{#'} in between the
-error-causing code and the \code{last_error()} or \code{last_trace()} call, to fulfill
-the requirement of being in separate chunks:
+To use \code{\link[rlang:last_error]{rlang::last_error()}} or \code{\link[rlang:last_error]{rlang::last_trace()}} within a reprex,
+you must place them in a different "chunk" to the code that generates an
+error. The easiest way to do is to insert a line containing the special
+comment \verb{#'} after error-causing code:
 
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{f <- function() rlang::abort('foo')
 f()