diff --git a/R/C5_rules.R b/R/C5_rules.R
index 0f8cb53..59aaa54 100644
--- a/R/C5_rules.R
+++ b/R/C5_rules.R
@@ -46,6 +46,7 @@
 #'  turns this feature off (although it can be re-enabled using
 #'  [C50::C5.0Control()]).
 #'
+#' @return An updated `parsnip` model specification.
 #' @seealso [parsnip::fit()], [parsnip::fit_xy()], [C50::C5.0()],
 #' [C50::C5.0Control()]
 #' @references Quinlan R (1993). _C4.5: Programs for Machine Learning_. Morgan
@@ -295,6 +296,10 @@ c5_pred <- function(object, new_data, trials = object$fit$trials["Actual"], ...)
 #' @param type A single character value or `NULL`. Possible values
 #'  are class" and "prob".
 #' @param ... Not currently used.
+#' @return A tibble with one row for each row of `new_data`. Multiple
+#' predictions are contained in a list column called `.pred`. That column has
+#' the standard `parsnip` prediction column names as well as the column with
+#' the tuning parameter values.
 #' @details
 #' For C5.0 rule-based models, the model fit may contain less boosting
 #' iterations than the number requested. Printing the object will show how many
diff --git a/R/cubist.R b/R/cubist.R
index 0fd4d0c..01fc143 100644
--- a/R/cubist.R
+++ b/R/cubist.R
@@ -63,6 +63,7 @@
 #'
 #' The only available engine is `"Cubist"`.
 #'
+#' @return An updated `parsnip` model specification.
 #' @seealso [parsnip::fit()], [parsnip::fit_xy()], [Cubist::cubist()],
 #' [Cubist::cubistControl()]
 #' @references Quinlan R (1992). "Learning with Continuous Classes." Proceedings
@@ -404,7 +405,7 @@ tunable.cubist_rules <- function(x, ...) {
   )
 }
 
-#' [multi_predict()] methods for rule-based models
+#' `multi_predict()` methods for rule-based models
 #' @rdname multi_predict
 #' @export
 #' @param neighbors An numeric vector of neighbors values between zero and nine.
diff --git a/R/rule_fit.R b/R/rule_fit.R
index f6176b4..a04407c 100644
--- a/R/rule_fit.R
+++ b/R/rule_fit.R
@@ -52,6 +52,7 @@
 #'
 #' The only available engine is `"xrf"`.
 #'
+#' @return An updated `parsnip` model specification.
 #' @seealso [parsnip::fit()], [parsnip::fit_xy()], [xrf::xrf.formula()]
 #' @references Friedman, J. H., and Popescu, B. E. (2008). "Predictive learning
 #' via rule ensembles." _The Annals ofApplied Statistics_, 2(3), 916-954.
@@ -492,6 +493,8 @@ tunable.rule_fit <- function(x, ...) {
 #' Proportion of Randomly Selected Predictors
 #'
 #' @inheritParams committees
+#' @return A `dials` with classes "quant_param" and "param". The `range` element
+#' of the object is always converted to a list with elements "lower" and "upper".
 #' @export
 mtry_prop <- function(range = c(0.1, 1), trans = NULL)  {
   dials::new_quant_param(
diff --git a/man/C5_rules.Rd b/man/C5_rules.Rd
index 777ad16..3e83272 100644
--- a/man/C5_rules.Rd
+++ b/man/C5_rules.Rd
@@ -38,6 +38,9 @@ or replaced wholesale.}
 
 \item{...}{Not used for \code{update()}.}
 }
+\value{
+An updated \code{parsnip} model specification.
+}
 \description{
 \code{\link[=C5_rules]{C5_rules()}} is a way to generate a \emph{specification} of a model
 before fitting. The main arguments for the model are:
diff --git a/man/cubist_rules.Rd b/man/cubist_rules.Rd
index 7b1e467..9981fe4 100644
--- a/man/cubist_rules.Rd
+++ b/man/cubist_rules.Rd
@@ -46,6 +46,9 @@ or replaced wholesale.}
 
 \item{...}{Not used for \code{update()}.}
 }
+\value{
+An updated \code{parsnip} model specification.
+}
 \description{
 \code{\link[=cubist_rules]{cubist_rules()}} is a way to generate a \emph{specification} of a model
 before fitting. The main arguments for the model are:
diff --git a/man/mtry_prop.Rd b/man/mtry_prop.Rd
index 42bedf0..aba85c3 100644
--- a/man/mtry_prop.Rd
+++ b/man/mtry_prop.Rd
@@ -15,6 +15,10 @@ largest possible values, respectively.}
 the default is used which matches the units used in \code{range}. If no
 transformation, \code{NULL}.}
 }
+\value{
+A \code{dials} with classes "quant_param" and "param". The \code{range} element
+of the object is always converted to a list with elements "lower" and "upper".
+}
 \description{
 Proportion of Randomly Selected Predictors
 }
diff --git a/man/multi_predict.Rd b/man/multi_predict.Rd
index d7b0191..9f25cc0 100644
--- a/man/multi_predict.Rd
+++ b/man/multi_predict.Rd
@@ -4,7 +4,7 @@
 \alias{multi_predict._c5_rules}
 \alias{multi_predict._cubist}
 \alias{multi_predict._xrf}
-\title{\code{\link[=multi_predict]{multi_predict()}} methods for rule-based models}
+\title{\code{multi_predict()} methods for rule-based models}
 \usage{
 \method{multi_predict}{`_c5_rules`}(object, new_data, type = NULL, trees = NULL, ...)
 
@@ -28,8 +28,14 @@ are class" and "prob".}
 
 \item{penalty}{Non-negative penalty values.}
 }
+\value{
+A tibble with one row for each row of \code{new_data}. Multiple
+predictions are contained in a list column called \code{.pred}. That column has
+the standard \code{parsnip} prediction column names as well as the column with
+the tuning parameter values.
+}
 \description{
-\code{\link[=multi_predict]{multi_predict()}} methods for rule-based models
+\code{multi_predict()} methods for rule-based models
 }
 \details{
 For C5.0 rule-based models, the model fit may contain less boosting
diff --git a/man/rule_fit.Rd b/man/rule_fit.Rd
index 9d4ef88..ad62936 100644
--- a/man/rule_fit.Rd
+++ b/man/rule_fit.Rd
@@ -72,6 +72,9 @@ or replaced wholesale.}
 
 \item{...}{Not used for \code{update()}.}
 }
+\value{
+An updated \code{parsnip} model specification.
+}
 \description{
 \code{\link[=rule_fit]{rule_fit()}} is a way to generate a \emph{specification} of a model
 before fitting. The main arguments for the model are: